Allosteric Network in Ube2t Drives Specificity for RING E3 Catalysed

bioRxiv preprint doi: https://doi.org/10.1101/429076; this version posted September 27, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1 Viduth K Chaugule1, 2*, Connor Arkinson1, 2, Rachel Toth2 and Helen Walden1, 2*

2 1Institute of Molecular, Cell and Systems Biology, College of Medical, Veterinary and Life

3 Sciences, University of Glasgow, Glasgow, UK

4 2MRC Protein Phosphorylation and Ubiquitylation Unit, College of Life Sciences

5 University of Dundee, Dundee, UK

6 * Corresponding authors. E-mail: [email protected],

7 [email protected]

9 Running title: E2 allostery for specific ubiquitination

10 Abstract

11 In eukaryotes, DNA damage repair is implemented by a host of proteins that are coordinated

12 by defined molecular signals. One such signal that transpires during the Fanconi Anemia

13 (FA) - interstrand crosslink (ICL) repair pathway is the site-specific monoubiquitination of

14 FANCD2 and FANCI proteins by a large, multi-protein FA core complex. The mechanics for

15 this exquisitely specific monoubiquitin signal has been elusive. Here we show FANCL, the

16 RING E3 module of the FA core complex, allosterically activates its cognate E2 Ube2T for

17 monoubiquitination by a mechanism distinct from the typical RING-based catalysis. FANCL

18 triggers intricate re-wiring of Ube2T’s intra-residue network thus activating the E2 for

19 precision targeting. This network is intrinsically regulated by conserved gates and loops

20 which can be engineered to yield Ube2T variants that enhance FANCD2 ubiquitination by

21 ~30-fold without compromising on target specificity. Finally, we also uncover allosteric

22 networks in other ubiquitin E2s that can be leveraged by RING E3 ligases to drive specific

23 ubiquitination.

26 Keywords: DNA repair / E2 / Enzyme allostery / RING E3 / Ubiquitination

27 Introduction

28 Ubiquitination is an essential, versatile and reversible post-translational modification system

29 that enables eukaryotic cells to reprogram the fate and function of the modified protein and

30 its connected pathway. The modification is accomplished by a sequential enzyme cascade

31 wherein ubiquitin’s C-terminus is first activated by an E1, is transferred onto the catalytic

32 cysteine of an E2 conjugase (E2~Ub) and finally E3 ligases mediate the covalent attachment

33 of ubiquitin onto a target lysine residue (Hochstrasser, 2009; Pickart, 2001). The Really

34 Interesting New Gene (RING) ligase proteins represent the largest E3 family (~600 members)

35 which share a zinc coordinating cross-brace motif termed RING domain (Freemont, 2000).

36 Mechanistically, while non-RING elements of E3 ligases specify the substrate, the RING

37 domains bind an E2 surface distal from the E2 active site and indirectly induce substrate

38 ubiquitination by stabilizing a productive E2~Ub conformation (Metzger et al., 2014).

39 Moreover, any of the seven surface lysine residues on ubiquitin or its N-terminus can be

40 targeted to build polyubiquitin chains, thus enabling diverse signals (Kulathu and Komander,

41 2012). Typically, RING-E2 interactions are found to be transient thus allowing E3s to switch

42 their E2 partners in order to assemble polyubiquitin signals on the substrate (Brown et al.,

43 2014; Kelly et al., 2014; Rodrigo-Brenni and Morgan, 2007; Windheim et al., 2008). Around

44 35 ubiquitin E2s are found in mammals, several of which build chains (Stewart et al., 2016).

45 Mechanisms of chain-assembly are well understood and generally involve additional non-

46 covalent interactions between the E2 and the acceptor ubiquitin surface proximal to the

47 linkage site (Eddins et al., 2006; Liu et al., 2014; Middleton and Day, 2015; Petroski and

48 Deshaies, 2005; Rodrigo-Brenni et al., 2010; Wickliffe et al., 2011). In contrast however, far

49 less is known about how RING E3-E2 enzyme pairs attach a single ubiquitin directly on the

50 substrate surface in the case of monoubiquitination.

51 Site-specific monoubiquitin signals feature prominently in fundamental DNA damage

52 response pathways (Al-Hakim et al., 2010; Uckelmann and Sixma, 2017). In eukaryotes, the

53 repair of toxic DNA interstrand cross-links (ICL) is mediated by the Fanconi Anemia (FA)

54 pathway, defects in which give rise to FA, a genome instability disorder typified by bone

55 marrow failure and high predisposition to cancers (Garaycoechea and Patel, 2014; Kottemann

56 and Smogorzewska, 2013). A key event in FA-ICL damage response is the site-specific

57 mono-ubiquitination of two large (~160kDa) structurally homologous proteins, FANCD2

58 (Garcia-Higuera et al., 2001) and FANCI (Sims et al., 2007; Smogorzewska et al., 2007)

59 (Lys561 and Lys523 respectively in humans), that signals the recruitment of repair factors

60 (Ceccaldi et al., 2016). The specific modification is mediated by the RING bearing protein

61 FANCL, present within a nine-protein FA core-complex E3 ligase (FANCA-FANCG-

62 FAAP20-FANCC-FANCE-FANCF-FANCB-FANCL-FAAP100) and the E2 Ube2T

63 (Machida et al., 2006; Meetei et al., 2003; Walden and Deans, 2014). FANCL’s central

64 region, a bi-lobed UBC (Ubiquitin conjugation fold)-RWD domain, facilitates direct

65 FANCD2/FANCI interaction (Cole et al., 2010; Hodson et al., 2011) while the C-terminal

66 RING domain selectively binds Ube2T over other E2’s (Hodson et al., 2014). Further,

67 genetic mutations in Ube2T, a Class III E2 with an unstructured C-terminal extension, have

68 recently been linked to a FA phenotype (Hira et al., 2015; Rickman et al., 2015; Virts et al.,

69 2015). Accordingly, in in vitro assays, the isolated FANCL and Ube2T enzymes catalyse

70 FANCD2 monoubiquitination, although the modification levels are unexpectedly low (Alpi et

71 al., 2008; Hodson et al., 2014). Studies in frog egg extracts show majority of FANCD2 is

72 present in complex with FANCI (Sareen et al., 2012) while the cell based data indicate mono-

73 ubiquitination of either protein requires the presence of the partner (Alpi and Patel, 2009).

74 However, a crystal structure of the mouse FANCI-FANCD2 complex reveals an extended

75 heterodimer interface which buries the respective target lysine (Joo et al., 2011). Notably, the

76 addition of structured or duplex DNA greatly stimulates FANCD2 ubiquitination and requires

77 the presence of FANCI (Longerich et al., 2014; Sato et al., 2012). The DNA binding

78 propensity of the FANCI-FANCD2 complex, absent in FANCL or Ube2T, is proposed to

79 induce local reconfiguration that could improve FANCL’s access to the target sites. Finally,

80 while FANCL alone induces low levels of ubiquitination, however when in a FANCB-

81 FANCL-FAAP100 sub-complex FANCD2 monoubiquitination levels improve by around 5-

82 fold. The added presence of a FANCC-FANCE-FANCF sub-complex progressively enhances

83 the tandem mono-ubiquitination of the FANCD2-FANCI complex (Rajendra et al., 2014; van

84 Twest et al., 2017). Thus, in the current model the presence of DNA, FANCI and additional

85 FA sub-complexes are all required for the modification however, underlying mechanisms for

86 the sub-complex induced enhancement are not well understood. First, FANCL dependent

87 FANCD2 ubiquitination has been observed in non-vertebrate species which lack an intact FA

88 core-complex, suggesting in part that the mechanism for the specific ubiquitination is

89 encoded within FANCL (Sugahara et al., 2012; Zhang et al., 2009). Second, global proteomic

90 profiling have uncovered several other lysine on human FANCD2 (22 sites) and FANCI (44

91 sites) that are ubiquitinated in vivo indicating the surface of both proteins are viable acceptors

92 (Kim et al., 2011; Udeshi et al., 2013). As ubiquitin signalling is influential in almost every

93 cellular process in eukaryotes, a major unanswered question is how specific signals are

94 assembled and regulated. The FA-ICL repair pathway is crucial for cellular homeostasis thus,

95 understanding how FANCL targets precise FANCD2 and FANCI sites for strict

96 monoubiquitination would provide valuable insights into the mechanics of this crucial DNA

97 damage response signal as well as how target and signal specificity is achieved in

98 ubiquitination.

99 In this study we show that FANCL activates Ube2T for ubiquitination through an allosteric

100 mechanism that is distinct from the typical RING E3 based catalysis. We find that in addition

101 to the selective FANCL RING-Ube2T interface, there are multiple E2-E3 interactions that

102 perturb the resting state of Ube2T to induce activity. Residue network analysis reveals subtle

103 reconfigurations of Ube2T’s internal connections that link the effect of FANCL binding to

104 the E2’s catalytic centre culminating in substrate ubiquitination. We further uncover intrinsic

105 regulation of this network by conserved Ube2T residues, and through rationally designed

106 mutations we can enhance FANCL mediated FANCD2 (~30 fold) and FANCI (~16 fold)

107 mono-ubiquitination without compromising its specificity. Finally, we identify similar

108 allosteric networks in other ubiquitin E2s that are appropriated by RING E3 ligases to drive

109 specific ubiquitination events.

110

111 Results

112 The E2 – E3 pair Ube2T – FANCL ubiquitinates its substrates via an atypical

113 mechanism

114 Previous in vitro studies using recombinant chicken (Alpi et al., 2008; Sato et al., 2012), frog

115 (Hodson et al., 2014) and human (Longerich et al., 2014) proteins report the isolated FANCL

116 enzyme with Ube2T directs FANCD2 monoubiquitination at its physiological target site.

117 However, the underlying mechanism for the site-specific and strict mono-modification is

118 unclear. In order to understand how the Ube2T and FANCL enzyme pair catalyse this

119 specific signal we reconstituted a minimal E2 – E3 module with human proteins. Based on

120 our earlier work we designed and purified a FANCL URD-RING fragment (FANCLUR,

121 residues 109-375) that is stable, monomeric (Supplementary Fig 1A) and comprises both the

122 substrate (UBC-RWD domain) and the E2 (RING domain) binding regions (Hodson et al.,

123 2011; Hodson et al., 2014). We then tested the activity of the FANCLUR fragment in in vitro

124 FANCD2 ubiquitination assays using fluorescently labelled ubiquitin (UbIR800). Previous

125 studies have shown the additional requirements of FANCI and DNA in complex with

126 FANCD2 for efficient monoubiquitination of this substrate. Consistent with this we observe

127 Ube2T – FANCLUR mediated FANCD2 modification when present as a FANCD2-FANCI-

128 dsDNA complex (Fig 1A). Further, an arginine mutant of the physiological FANCD2 target

129 site (FANCD2 K561R) prevents ubiquitination thus confirming the minimal E2 – E3 module

130 is both active and site-specific. We wondered if the FANCLUR fragment could also

131 specifically ubiquitinate FANCI present in the FANCD2-FANCI-dsDNA complex. To test

132 this we titrated increasing amounts of the E2 – E3 module in reactions containing single

133 target site complexes (FANCIK523R-FANCD2 or FANCI-FANCD2K561R). FANCLUR clearly

134 favours modification of the FANCD2 site (Fig 1B). In contrast, FANCLUR can drive the site-

135 specific ubiquitination of an isolated FANCI-dsDNA complex (Fig 1C), as previously

136 observed (Longerich et al., 2014). Therefore, unless otherwise stated, all FANCD2 assays are

137 in the presence of FANCI and dsDNA, while FANCI assays are in the presence of dsDNA.

138 RING E3 ligases catalyse ubiquitination by activating the thioester linked E2~ubiquitin (Ub)

139 intermediate. In brief, RING binding of E2~Ub induces the thioester linked Ub to fold back

140 over the E2 wherein the Ile44-centered hydrophobic patch of ubiquitin packs against a central

141 E2 helix. Concomitantly, a ‘linchpin’ RING E3 residue (usually Arg/Lys) contacts both E2

142 and Ub to stabilise the ‘closed’ E2~Ub conformer, thus priming the thioester for lysine attack

143 (Dou et al., 2012; Plechanovova et al., 2012; Pruneda et al., 2012; Saha et al., 2011).

144 Interestingly, FANCL lacks this linchpin harbouring instead a serine residue (S363) at the

145 analogous position (Supplementary Fig 1B). We therefore wondered if in the absence of a

146 linchpin residue the Ub Ile44 patch requirement is maintained for FANCL’s E3 activity. To

147 test this we compared FANCI and FANCD2 ubiquitination activity of the Ube2T –

148 FANCLUR pair with the well characterised E2 – E3 pair Ube2D3 – RNF4RING-RING (Branigan

149 et al., 2015; Plechanovova et al., 2011; Plechanovova et al., 2012). While a FANCI-

150 FANCD2-DNA complex is robustly ubiquitinated by Ube2D3 – RNF4RING-RING, the Ile44Ala

151 Ub mutant dramatically reduces this modification (Fig 1D). Remarkably, the ubiquitin mutant

152 barely impacts the activity or site-specificity of the Ube2T-FANCLUR pair. These data reveal

153 that while Ube2T–FANCL can catalyse specific ubiquitination it does not share features of

154 the generic RING E3 based catalysis and instead operates through an as yet uncharacterised

155 mechanism.

156

157 FANCL stimulates Ube2T activity through allosteric modulation

158 As the minimal Ube2T–FANCLUR module drives FANCI and FANCD2 monoubiquitination,

159 we wondered if the underlying mechanism for specific ubiquitination could be uncovered by

160 understanding FANCL’s atypical catalytic mechanism. To investigate this we compared

161 available structures of unbound and the FANCL bound E2. High resolution crystal structures

162 of Ube2T alone (PDB ID 1yh2) (Sheng et al., 2012) and bound to FANCL RING domain

163 (FANCLR, PDB ID 4ccg) (Hodson et al., 2014) show little overall difference in their UBC

164 folds (residues 1-152). The latter contains two E2 copies in the asymmetric unit, both of

165 which superpose well onto the unbound state (RMS deviation 0.6-0.9Å). However, RING

166 binding induces some local changes in Ube2T’s helix1–loop2 region as well as in loops 7 and

167 8 that flank the active site (Fig 2A). Residues in these shifted regions (R3, L7, D32, D33,

168 K91-K95 and D122) are predominantly surface exposed and largely conserved among the

169 Ube2T homologs (Supplementary Fig 1C). We wondered if these subtle conformational

170 changes are important for FANCL’s catalytic mechanism. To test this we made and assayed

171 Ube2T mutants in FANCI and FANCD2 ubiquitination assays. Surprisingly, alanine

172 substitutions of residues Arg3, Asp32/Asp33 and Leu92 reduce the rate of

173 monoubiquitination by 25 to 50% while a loop7 deletion (∆92-95) causes a more severe

174 defect (Fig 2B-C and Supplementary Fig2A). Interestingly, a loop2-loop7 hybrid mutant

175 (D32A, D33A, L92A or DDL/AAA) attenuates both FANCI and FANCD2 ubiquitination

176 rates by around 75%. It is possible that these substrate ubiquitination defects arise from the

177 mutations impairing intrinsic E2 features, such as ubiquitin charging and discharging. The

178 levels of E1-based Ub charging for the above Ube2T mutants is similar to the wildtype E2

179 (Supplementary Fig 2B). Ube2T readily autoubiquitinates at Lys91, close to the active site

180 (Fig 2A), and several lysines in its C-terminal extension (Machida et al., 2006). We therefore

181 purified an E2 truncation (Ube2T1-152) lacking the C-terminal tail. This truncated enzyme

182 targets Lys91 and can be used to assess E2~Ub discharge. In E3-independent Ube2T1-152

183 autoubiquitination assays, a Lys91Arg mutation (Ube2T1-152 K91R) abolishes the

184 automodification while the DDL/AAA mutation (Ube2T1-152 DDL/AAA) has no observable

185 effect (Fig 2D). Thus in the absence of FANCL, the Ube2TDDL/AAA mutant can load and

186 offload ubiquitin comparable to wildtype Ube2T. Conversely, in single turnover reactions,

187 the mutant E2~Ub thioester (Ube2T1-152, K91R DDL/AAA ~ Ub) is less effective in FANCLUR

188 mediated FANCD2 ubiquitination even in the presence of increasing amounts of the E3 (Fig

189 2E).

190 We wondered if this disparity arises from FANCL being sensed differently by the mutant

191 Ube2T. To uncover possible differences we analysed interactions of wildtype and

192 DDL/AAA mutant Ube2T with the FANCLUR fragment in solution and observe similar

193 affinities (Kd ~119 nM) indicating that the RING-Ube2T crystal interface is maintained in

194 case of the Ube2TDDL/AAA (Fig 2F). However, the thermodynamics are fundamentally

195 different as the mutant interaction is enthalpically favoured (ΔH = -1.65 kcal/mol) in contrast

196 to the unfavourable signature observed for wildtype Ube2T (ΔH = + 6.46 kcal/mol) (Fig 2F,

197 Table 1). The shorter FANCLR fragment also binds with ~250 nM affinity but shows a

198 similar divergent enthalpy profile (Supplementary Fig 2C). In both FANCLUR and FANCLR

199 experimental sets there appears to be strong enthalpy – entropy compensation and

200 consequently the binding energy (ΔG) within each set is unchanged (Table 1). Therefore, the

201 net differences in observed entropy between wildtype and DDL/AAA Ube2T (8.12 and 6.73

202 kcal/mol for FANCLUR and FANCLR complexes respectively) could arise from either

203 reduced solvent reordering or fewer local conformational changes in the mutant E2 upon

204 FANCL binding. In other words, the Ube2TDDL/AAA mutant indeed senses FANCL differently

205 from wildtype Ube2T. The biochemical and thermodynamic data taken together reveal that

206 FANCL induces FANCI and FANCD2 ubiquitination by perturbing the overall resting state

207 of Ube2T and strongly suggests allostery. Moreover, FANCL binding indirectly effects

208 Ube2T’s loop2 and loop7 which are 16 – 20Å apart (Fig 2A) and these regions, operating in

209 synergy, propagate the catalytic influence of the E3.

210

211 Ube2T backside regulates FANCL mediated FANCD2 ubiquitination

212 Our binding analyses reveal a 2-fold enhanced Ube2T affinity for the FANCLUR fragment

213 over the smaller RING domain (Table 1, Supplementary Fig 2C) indicating additional

214 interactions exist between a non-RING element and the E2. In several RING E3s, auxiliary

215 elements outside of the RING domain can modulate ubiquitination by binding an E2

216 ‘backside’ surface which is located opposite the active site (Brown et al., 2015; Das et al.,

217 2009; Hibbert et al., 2011; Li et al., 2015; Li et al., 2009; Metzger et al., 2013; Turco et al.,

218 2015). We wondered if the analogous Ube2T backside surface could extend the FANCL-E2

219 interface as well as influence FANCI and FANCD2 ubiquitination. Beta-strands 1 and 2 of

220 the UBC-fold prominently feature in E3 – backside E2 complexes (PDB IDs 3h8k, 2ybf,

221 4jqu, 5d1k and 4yii) and at the canonical ubiquitin – backside E2 interface (PDB IDs 2fuh

222 and 4v3l). The equivalent Ube2T surface is hydrophobic, semi-conserved (Supplementary

223 Fig 1C) with certain side-chains (β1 – T23, W25 and β2 – R35, Q37) repositioned upon

224 FANCLR binding (Fig 3A). Notably, the mutation of Ube2T β1 (T23R+W25Q or TW/RQ)

UR R 225 reduces the E2’s affinity for FANCL (Kd – 200 nM) but not for FANCL (Fig 3B, Table

226 1). Thus, the Ube2T backside indeed supports additional interactions with FANCLUR beyond

227 the RING domain. Unexpectedly, the backside mutants have different effects on substrate

228 ubiquitination. Previous in vitro studies have reported site-specific ubiquitination of a

229 FANCI/DNA complex by Ube2T in the absence of FANCL (Longerich et al., 2014). In our

230 setup, FANCLUR enhances FANCI ubiquitination rates by around two-fold while backside

231 Ube2T mutants mitigate this improvement (Fig 3C-D and Supplementary Fig 3A). In

232 contrast, the Ube2T TW/RQ mutant slows FANCD2 ubiquitination by over three-fold,

233 similar to the DDL/AAA mutant (Fig 2B-C). We also tested a β2 mutant (Q37L), designed to

234 extend the hydrophobic backside surface, and observe little change in ubiquitination rates

235 (Fig 3C, D). Overall, the backside Ube2T mutants do not affect ubiquitin charging or

236 Ube2T1-152 auto-ubiquitination (Supplementary Fig 3A and 3E) and therefore indicate the

237 ubiquitin loading/offloading properties of the mutants are intact. Hence, the observed defects

238 in substrate modification are likely linked to an altered FANCLUR-backside Ube2T

239 interaction. We wondered if the weakened Ube2T TW/RQ-FANCLUR interaction is solely

240 responsible for reduced FANCD2 ubiquitination. In single-turnover reactions, the charged

241 Ube2T1-152, K91R TW/RQ ~ Ub thioester is weakly activated by FANCLUR for FANCD2

242 ubiquitination however, increasing the amount of E3 does not completely rescue the defect

243 (Fig 3F). Thus, Ube2T’s backside surface not only supports FANCLUR interaction but likely

244 augments the allosteric activation of the E2~Ub thioester by the E3. In summary, loop2,

245 loop7 and the backside of Ube2T together respond to FANCL binding to facilitate FANCI

246 and FANCD2 ubiquitination.

247

248 FANCL potentiates Ube2T active site residues for FANCI AND FANCD2

249 ubiquitination

250 The above data reveal how the FANCL influence on distinct Ube2T surfaces triggers the

251 E2~Ub for substrate ubiquitination. Thus a long-range residue network could connect these

252 distal sites with the E2 catalytic centre. To uncover the likely path we generated residue

253 interaction networks (RINs) for the free and RING bound Ube2T structures (residues 1-152)

254 (Doncheva et al., 2011; Piovesan et al., 2016). In these networks, each Ube2T residue is

255 represented as a node while the connecting edges are potential physicochemical interactions

256 with its tertiary structure environment. The total connections in both free and RING bound

257 E2 RINs are similar, averaging 1430 edges, however a comparison matrix reveals unchanged

258 and altered edges (Fig 4A, Supplementary Dataset 1), the latter used to build a dynamic

259 network. We then choose Phe63, a core E2 residue at the Ube2T-FANCLR interface (Hodson

260 et al., 2014), as the starting node to trace its first neighbours which serve as subsequent

261 search nodes. By iteration, we trace the possible paths to the E2 catalytic centre, focusing on

262 the allosteric and conserved nodes while filtering out paths comprising distal and dead-end

263 nodes. The final allosteric network model (39 nodes, 79 dynamic edges) reveals how

264 FANCLR binding rewires Ube2T’s intra-molecular connections (Fig 4A, Supplementary

265 Table 1).

266 The network terminals, located in the catalytic beta-element (R84), loop7 (K91, K95), loop8

267 and its C-terminal hinge (D122 and L124 respectively) are within 10Å of Ube2T’s catalytic

268 cysteine (C86) and vary among the ubiquitin E2s (Fig 4A and Supplementary Fig 4B). To

269 empirically test the network model we made alanine mutants of the said network termini and

270 observe a striking loss in FANCL mediated FANCI and FANCD2 ubiquitination (Fig 4B).

271 Given their proximity to the active site these Ube2T residues could also influence intrinsic E2

272 activity. The Leu124Ala mutation is detrimental to E1-based ubiquitin charging and suggests

273 the hydrophobic side-chain braces Ube2T’s active site for optimal activity (Fig 4C).

274 Moreover, by examining E3-independent E2~Ub thioester discharge onto free lysine, we

275 observe subtle catalytic defects with the loop7 (Ube2T1-152 K91A+K95A) and loop8 (Ube2T1-

276 152, K91R D122A) mutants while in contrast, the Arg84Ser (Ube2T1-152, K91R R84S) mutant did

277 not affect Ube2T’s aminolysis activity (Fig 4D and Fig Supplementary Fig 4C). In some E2s,

278 the loop8 acidic residue is required to position and/or deprotonate the target lysine for

279 modification (Plechanovova et al., 2012; Valimberti et al., 2015; Yunus and Lima, 2006).

280 The loop8 Asp122 in Ube2T could support a similar role and account for the catalytic defects

281 with this mutant. However, the contrasting E2 activity profiles (E3-independent free lysine

282 versus E3-dependent substrate lysine, Fig 4B and D) for the catalytic beta-element mutant

283 (R84S) suggests the Arg84 residue is likely involved in FANCL’s activation mechanism.

284 Moreover, as we earlier observe the Leu92Ala loop7 mutant of Ube2T reduces FANCL

285 mediated substrate ubiquitination (Fig 2B-C) we reasoned the E3’s activation mechanism

286 feeds into the catalytic role of this loop (Fig 4A, D).

287 To understand roles of Ube2T residues Arg84, Lys91 and Lys95 in substrate ubiquitination

288 we undertook systematic mutagenesis and uncover the requirements of Lys/Arg in loop7

289 while the Arg84 residue is indispensable for FANCL mediated FANCI and FANCD2

290 ubiquitination (Fig 4E). To clarify this necessity we analysed the total network (unchanged

291 and altered) for the Arg84 node (Supplementary Fig 4D). In free Ube2T, the Arg84 side-

292 chain stabilises the Asp80-Gly83 loop and is transiently redirected towards Cys86 upon

293 RING binding (Fig 4A, Supplementary Fig 4B and 4D). A Arg84Lys mutation would, in

294 theory, preserve the free E2 network but the persistent defect in substrate ubiquitination (Fig

295 4E) suggests the shorter Lys side-chain is unable to optimally participate at the active site.

296 Interestingly, the introduction of longer Arg side-chains in loop7 (R84K, K91R, K95R) can

297 rescue the Arg84Lys defect, albeit partially (Fig 4E). In summary, FANCL binding

298 potentiates the Ube2T active site residues R84, K91 and K95 to induce FANCI and FANCD2

299 ubiquitination. These residues could enhance catalysis either through transient interactions

300 with acidic/polar substrate residues proximal (<12Å) to the respective target lysine

301 (Supplementary Fig 4C) and/or stabilise the developing oxyanion hole in the E2~Ub – target

302 lysine transition state. Taken together, the data reveals the existence of an elaborate Ube2T

303 residue network that propagates FANCL’s catalytic influence in activating the E2~Ub

304 thioester for ubiquitination.

305

306 Ube2T network analysis reveals regulatory and FANCL induced activation mechanisms

307 The thermodynamics of Ube2T-FANCL interaction and the allosteric network model together

308 illustrate how the E3 could remotely effect the E2 catalytic centre by inducing a series of

309 subtle conformational changes within Ube2T. Intriguingly, we noticed a dynamic conduit

310 involving nodes Arg35 – Glu54 – Arg69 that appear to propagate FANCL’s influence on the

311 RING and backside clusters to the catalytic beta-element (Fig 4A, Supplementary Table 1).

312 In particular, the Glu54 side-chain which engages both Arg35 and Arg69 in unbound Ube2T

313 has a reduced influence in the RING bound state. Subsequently, the Arg69 side-chain is

314 liberated to stabilise the catalytic beta-element backbone thus releasing the Arg84 side-chain

315 to optimally participate at the catalytic centre (Fig 5A). In other words, an Arg69 effector role

316 in Ube2T’s resting state could be gated by Glu54 and relieved upon FANCL binding thereby

317 activating the E2~Ub for FANCI and FANCD2 ubiquitination. Based on this model we

318 predict the removal of residue 54’s acidic side-chain should positively impact substrate

319 ubiquitination. We tested the proposed allosteric conduit using Glu54Ala/Gln mutations and

320 observe a marked improvement in FANCLUR mediated FANCI AND FANCD2

321 ubiquitination while a conservative Glu54Asp mutant retained wild-type like activity (Fig

322 5B). Even a charge-altering Glu54Arg mutation contributes to greater substrate ubiquitination

323 and does not influence the Ube2T-FANCLUR interaction (Fig 5B, Table 1). Consequently, an

324 Arg69Ala mutation reduces FANCI and FANCD2 ubiquitination, while a conservative

325 mutation (R69K) rescues this defect and is further improved by eliminating the gating effect

326 (E54A with R69K) thus confirming residue 69’s effector role (Fig 5C). Furthermore,

327 disrupting the Glu54 gate can rescue FANCD2 ubiquitination defects arising from both RING

328 allostery (DDL/AAA) and backside binding (TW/RQ) mutants (Fig 5D). Interestingly, a

329 RING binding E2 mutant (F63A) that reduces Ube2T-FANCLR binding by over ten-fold

330 (Hodson et al., 2011) could also be activated by FANCLUR for FANCD2 ubiquitination by

331 using a permissive gate (F63A with E54R) (Fig 5D).

332 We wondered if such gated networks exists in other E2s, in particular those that assemble

333 specific ubiquitination signals. For example, DNA damage tolerance pathways are initiated

334 upon specific Lys164 monoubiquitination of Proliferating Cell Nuclear Antigen (PCNA)

335 protein by the E2-E3 enzyme pair Ube2B/Rad6-RAD18 (Ulrich and Walden, 2010). The

336 RAD18-Ube2B interaction is bimodal, primarily via the canonical RING-E2 interface and

337 supported by a helical Rad6 binding domain (R6BD) that packs against the E2’s backside

338 (Hibbert et al., 2011; Huang et al., 2011). We compared the R6BD bound (PDB ID 2ybf)

339 structure/RIN with the free Ube2B (PDB ID 2yb6) and noticed Glu58 and Arg71 of the E2

340 could potentially operate as gating and effector residues respectively (Supplementary Fig 5A,

341 Supplementary Dataset 2). Notably in in vitro assays, the Ube2B variant with a permissive

342 gate (E58R) is more sensitive to Rad18 in PCNA monoubiquitination without comprising

343 site-specificity (Supplementary Fig 5A). These results together demonstrate that different

344 E2’s contain long-range RINs that are regulated by internal gates. Furthermore, these

345 networks can be leveraged by RING E3 ligases to allosterically drive ubiquitination of

346 specific targets.

347 A short-range RIN is also apparent between Ube2T residues that support RING binding and

348 those in loop7 (Fig 4A, Supplementary Table 1). The binding of FANCLR draws loop7

349 (K91LPPK95) away from the active site (Fig 2A) thereby repositioning the flanking Lys

350 residues (K91 and K95) presumably for target site binding, while paradoxically these

351 residues are also needed at the active site for optimal Ube2T catalysis (Fig 4C). Also present

352 in this loop are two conserved proline residues (P93 and P94, Supplementary Fig 1C) that

353 maintain loop rigidity and we hypothesise that a plastic loop7 could emulate its dynamic

354 requirements during ubiquitination. In multi-turnover assays, the Ube2Tv1 variant with a

355 permissive gate (E54R) improves FANCLUR driven FANCD2 ubiquitination by around 15-

356 fold (around 4-fold for FANCI) while a hybrid Ube2Tv3 variant that includes both a

357 permissive gate and flexible loop7 (E54R, P93G, P94G) further enhances activity by 2 to 4-

358 fold (Fig 5E and Supplementary Fig5B) thus validating our hypothesis. The above data reveal

359 distinct regulatory mechanisms in operation within Ube2T’s allosteric network and these can

360 be repurposed to yield E2 variants with enhanced catalytic potential. Furthermore, the

361 enhanced Ube2Tv3 – FANCLUR pair facilitates steady and site-specific ubiquitination of the

362 isolated FANCD2 and FANCI proteins in the absence of any DNA cofactors (Fig 5F). Recent

363 reports reveal the in-vitro requirements of a large (~ 0.8 MDa), multi-protein E3 super-

364 assembly (FANCB-FANCL-FAAP100-FANCC-FANCE-FANCF) that activates the

365 Ube2T~Ub thioester for robust ubiquitination of a FANCI-FANCD2-DNA complex (Swuec

366 et al., 2017; van Twest et al., 2017). By investigating the catalytic mechanism of the core E2

367 – E3 proteins within this super-assembly i.e. Ube2T – FANCL, we have rationally engineered

368 a minimal module (Ube2Tv3 – FANCLUR, ~ 0.05 MDa) with enhanced catalytic ability to

369 autonomously, specifically and efficiently mono-ubiquitinate FANCI or FANCD2. Using this

370 simplified setup we can now systematically examine roles of the FA core-complex members

371 in ubiquitination, biochemically and structurally characterize the natively ubiquitinated

372 FANCI and FANCD2 substrates, identify novel readers of the ubiquitin signal that implement

373 DNA repair as well as understand how this specific signal is removed by the cognate

374 deubiquitinating enzyme USP1 (Nijman et al., 2005).

375

376 Discussion

377 Understanding the mechanisms for site-specific FANCD2 and FANCI mono-ubiquitination

378 would shed light on how this decisive DNA damage response signal is mediated by the multi-

379 protein FA core-complex as well as divulge ubiquitination strategies used in precision

380 targeting. Our previous studies on FANCL, the catalytic RING-bearing subunit in the FA

381 core-complex, revealed substrate binding via its central UBC-RWD domain (Hodson et al.,

382 2011). Moreover, an extended RING-E2 interface underlies the strong E3-E2 affinity and

383 enables FANCL to specify its E2 partner Ube2T (Hodson et al., 2014). However, the

384 mechanism by which FANCL catalyses monoubiquitination at specific FANCD2 and FANCI

385 sites, a key signal in the FA-ICL repair pathway (Ceccaldi et al., 2016), is poorly understood.

386 In this study we expand the functional significance of the intimate FANCL-Ube2T interaction

387 and uncover an atypical mechanism behind substrate ubiquitination.

388 Unlike other RING E3s, FANCL lacks the conserved linchpin residue (Arg/Lys) essential for

389 stabilizing a closed and productive E2~Ub conformation (Metzger et al., 2014). This suggests

390 FANCL’s RING domain may not activate most E2~Ub intermediates thus precluding any

391 non-specific modification or the assembly polyubiquitin signals on the FANCI and FANCD2

392 substrates. Despite missing this feature FANCL does activate Ube2T for site-specific

393 substrate ubiquitination. Using thermodynamic and residue network analysis we demonstrate

394 how FANCL’s high-affinity grasp of Ube2T induces a series of subtle conformational

395 changes within the E2 that are relevant for substrate ubiquitination. These changes transpire

396 through altered intra-residue connections between conserved Ube2T residues, revealing a

397 dynamic allosteric network that links FANCL binding to the E2 catalytic centre. Notably,

398 conserved basic residues (Arg84, Lys91 and Lys95) proximal to Ube2T’s catalytic cysteine

399 (Cys86) are repositioned in the FANCLR bound structure. While these residues partially

400 contribute to intrinsic Ube2T activity, they are critical for FANCL induced FANCD2 and

401 FANCI ubiquitination. Several studies on site-specific histone monoubiquitination

402 mechanisms show how interactions between RING domain residues and the substrate surface

403 guide the catalytic RING-E2~Ub complex to a lysine targeting zone (Bentley et al., 2011;

404 Gallego et al., 2016; Mattiroli et al., 2014; Mattiroli et al., 2012; McGinty et al., 2014). The

405 FANCL-Ube2T complex is similar in that substrate docking by FANCL’s UBC-RWD

406 module could restrict the global targeting area of the RING bound Ube2T. However, unique

407 to this E3-E2 pair is FANCL RING allostery which reorients the basic residues near Ube2T’s

408 active-site, which could facilitate local contacts with conserved acidic/polar

409 FANCD2/FANCI residues in the vicinity of the target lysine, thus directing specific

410 ubiquitination. It remains to be seen if other RING E3s can actively repurpose E2 residues for

411 specific lysine targeting. However, this strategy of E2-guided lysine targeting mirrors those

412 observed in autonomous polyubiquitin assembling E2s where the acceptor ubiquitin surface

413 around the target lysine is homed by auxiliary E2 interactions (Eddins et al., 2006; Liu et al.,

414 2014; Middleton and Day, 2015; Petroski and Deshaies, 2005; Rodrigo-Brenni et al., 2010;

415 Wickliffe et al., 2011). Alternatively, given the necessity of DNA cofactors in

416 FANCI/FANCD2 modification (Liang et al., 2016; Longerich et al., 2014; Sato et al., 2012),

417 a DNA sensing role for these Ube2T basic residues cannot be ruled out. In either scenario,

418 post-ubiquitination the local target zone for the FANCL-Ube2T complex is occluded by the

419 attached mono-ubiquitin. As we do not observe continual modification, either on a different

420 substrate lysine or the installed ubiquitin, we propose neither surface is efficiently recognized

421 by the FANCL-Ube2T complex thus limiting the ubiquitination to a single event.

422 Further, we uncover residues on Ube2T’s backside (β1 and β2) that support additional

423 interactions with FANCL and feature in the E2’s allosteric network. It is likely that the

424 backside Ube2T binding is mediated by FANCL’s UBC-RWD domain which also docks onto

425 the substrate surface. As the backside interaction is required for steady substrate

426 ubiquitination, FANCD2 in particular, we speculate that additional UBC-RWD and Ube2T

427 interactions could guide the E2~Ub to the substrate as well as allosterically activate the

428 enzyme. Notably, these observations enlist Ube2T to the growing number of E2’s whose

429 ubiquitination activities are modulated by interactions outside of the classical RING-E2

430 interface (Brown et al., 2015; Das et al., 2009; Hibbert et al., 2011; Li et al., 2015; Li et al.,

431 2009; Metzger et al., 2013; Turco et al., 2015). Finally, the network guided biochemical

432 analysis also reveals the presence of dynamic long-range and short-range residue networks

433 that are intrinsically regulated by conserved Ube2T residues. In particular, an allosteric gating

434 residue in β3 and a rigid loop7 appear to regulate the intensity of Ube2T activation by

435 FANCL. Rationally engineered mutations of these regions give rise to deregulated Ube2T

436 variants which are strikingly more responsive to FANCL in substrate ubiquitination, yet

437 retain the lysine specificity observed with the wildtype E2. Notably, the catalytic enhanced

438 Ube2T variants now support FANCL driven monoubiquitination of the isolated FANCD2

439 and FANCI proteins without needing DNA cofactors. These data reveal the molecular

440 strategies in place in Ube2T that prevent inadvertent and untimely mono-ubiquitin signals in

441 the FA-ICL repair pathway. Taken together, we propose a model where FANCL binding at

442 the canonical RING-E2 interface and the backside E2 surface rewires Ube2T’s residue

443 network and thus activates the enzyme for site-specific FANCI and FANCD2

444 monoubiquitination (Fig. 5G).

445 On the basis of Ube2T’s allosteric network we identify a similar regulated network in

446 Ube2B/Rad6 that is altered by the RING E3 Rad18 for Lys164 PCNA mono-ubiquitination.

447 Biochemical, structural and computational studies have revealed how RING/U-box E3s

448 stimulate internal dynamics in different E2s that are linked to their ubiquitination activity

449 (Benirschke et al., 2010; Chakrabarti et al., 2017; Das et al., 2013; Ozkan et al., 2005).

450 Furthermore, recent efforts in fragment-based inhibitor discovery have revealed promising

451 lead compounds that bind Ube2T (Morreale et al., 2017) as well as the unrelated Ube2I

452 (Hewitt et al., 2016) at regions distal from their active/E3 binding sites, nevertheless can

453 allosterically regulate the respective E2 activities. These observations collectively suggest

454 that allosteric networks could operate across the E2 family and that future investigations into

455 such networks could prove instrumental for basic and translational research in ubiquitin

456 biology. Using our Ube2T-gated network as a guide and the available E2 sequence/structural

457 data, we propose at least 8 other ubiquitin E2s (Ube2- A, C, E1, E2, E3, K, L3 and N) contain

458 a β3 gating residue that is restrictive (Supplementary Fig 5C). In contrast, the small ubiquitin-

459 like modifier (SUMO) E2 Ube2I and Interferon-stimulated gene 15 (ISG15) E2 Ube2L6

460 appear to contain a permissive β3 gate. Interestingly, opposing allosteric gates are apparent

461 for the two neural precursor cell expressed developmentally downregulated protein 8

462 (NEDD8) E2s Ube2F (restrictive) and Ube2M (permissive). Since the β3 gating residue does

463 not lie in the typical RING or backside binding regions, we speculate that targeting this

464 residue would maintain E2-E3 interactions nevertheless, will yield E2 variants that differ in

465 their enzymatic potential relative to their wildtype counterparts. Recent studies have utilized

466 phage-display derived ubiquitin binding variants to uncover mechanistic insights for

467 RING/U-box (Gabrielsen et al., 2017) and HECT (Zhang et al., 2016) E3s as well as

468 deubiquitinating enzymes (Ernst et al., 2013). Along these lines, we propose the design

469 principles for creating E2 activity variants that can be used in fundamental research and guide

470 drug discovery studies.

471

472 Materials and Methods

473 Cloning and mutagenesis of expression constructs

474 Human Ube2T and Ube2D3 was cloned using PCR/restriction cloning into a modified pET15

475 vector (Novagen) that express with a 6xHis-3C cleavage site at the N-terminus. Human

476 Ube2B cDNA was as an I.M.A.G.E. clone (Geneservice) and cloned into a modified

477 pDEST17 vector (Invitrogen), that express with a 6xHis-TEV cleavage site at the N-terminus

478 using the Gateway Cloning kit (Invitrogen). A synthetic human FANCL sequence (GeneArt)

479 optimized for Escherichia coli (E. coli) expression used as template to clone the FANCLUR

480 (residues 109-375) and FANCLR (residues 289-375) coding regions into a pET SUMO

481 (Invitrogen) expression vector using restriction-free cloning (van den Ent and Lowe, 2006).

482 These constructs express with a 6xHis-smt3 tag at the N-terminus. A synthetic human Rad18

483 sequence (GeneArt) optimized for Escherichia coli expression was cloned using

484 PCR/restriction cloning into a modified pET28a (Novagen), that express with 6xHis-smt3 tag

485 at the N-terminus. A synthetic Human FANCD2 sequence (GeneArt) optimized for

486 Spodoptera frugiperda (Sf) expression was cloned using PCR/restriction cloning into a

487 pFastBac vector that express with a 6xHis-3C cleavage site at the N-terminus. Human

488 FANCI (cDNA purchased as an I.M.A.G.E. clone, Geneservice) was cloned using

489 PCR/restriction cloning into a pFastBac vector that express with a 6xHis-TEV cleavage site-

490 V5 epitope at the N-terminus. Human PCNA template, a kind gift from Dr Svend Petersen-

491 Mahrt (European Institute of Oncology, Milan), was cloned using PCR/restriction cloning

492 into a pRSF Duet1 vector (Novagen), that express with a 6xHis at the N-terminus. Ubiquitin

493 was cloned using PCR/restriction cloning into a modified pET28a vector (Novagen), that

494 express with 6xHis-smt3 tag-SGCGSG overhang at the N-terminus for fluorescence labelling

495 or into a modified pRSF Duet1 vector (Novagen), that express with a 6xHis-TEV cleavage

496 site at the N-terminus. All desired mutagenesis were carried out by primer based PCR

497 mutagenesis using KOD Hot Start DNA polymerase (Novagen) or Phusion High-Fidelity

498 DNA polymerase (Thermo Scientific) following manufacturer’s guidelines. Custom

499 oligonucleotides for PCR and mutagenesis were obtained from Sigma Aldrich or IDT

500 technologies. DH5α E. coli strain were made chemically competent using CCMB80 buffer

501 (TEKnova) using in-house protocols. The coding regions of all constructs and mutants were

502 verified by DNA sequencing using MRC PPU DNA Sequencing and Services or Eurofins

503 Genomics DNA sequencing services.

504 Expression and purification of recombinant proteins

505 All E2, Ubiquitin and PCNA constructs were transformed into chemically competent BL21

506 E. coli strains and cultured in Miller LB broth (Merck) at 37°C until OD600 ~ 0.4 following

507 which temperature was reduced to 16°C. At OD600 ~ 0.8 protein expression was induced by

508 adding a final concentration of 200 μM (for Ube2T and Ube2D3) or 500 μM (for Ube2B,

509 Ubiquitin and PCNA) Isopropyl-Beta-d-Thiogalactopyranoside (IPTG, Formedium) and

510 allowed to proceed for a further 16-18h. Similar steps were followed for all RING domain

511 constructs except the growth media was supplemented with 250 μM ZnCl2 and protein

512 expression was induced at OD600 ~ 1.0 with a final concentration of 50 μM IPTG. All

513 purification steps, except for ubiquitin, were performed at 4°C and completed within 24-36

514 hours of cell lysis.

515 Purification of all Ube2T variants and FANCL fragments are as described in (Hodson et al.,

516 2011; Hodson et al., 2014). The respective affinity tags were cleaved using GST-3C protease

517 (for Ube2T) and 6xHis-Ulp1 protease (for FANCL). Ube2B variants and PCNA were

518 expressed and purified similar to Ube2T. The 6xHis tag was removed from Ube2B using

519 6xHis-TEV protease. The 6xHis-smt3-Rad18 and 6xHis-MBP-rat RNF4 RING-RING

520 proteins was purified as described elsewhere (Huang et al., 2011; Plechanovova et al., 2011).

521 The 6xHis-smt3 was retained for RAD18 and protein concentration for both proteins were

522 estimated using SDS-PAGE and coomassie staining using known protein standards of similar

523 size.

524 Baculovirus were generated using the Bac-to-Bac (Invitrogen) system and proteins were

525 expressed for ~72h in baclovirus infected Sf21 cells. Cell pellets were suspended in lysis

526 buffer (50 mM Tris-HCl pH8.0, 150 mM NaCl, 10 mM imidazole, 1 mM TCEP and 5% v/v

527 glycerol with freshly added 2 mM MgCl2, EDTA-free Protease Inhibitor cocktail (Pierce) and

528 Benzonase (Sigma Aldrich). Sf21 cells were lysed by homogeniser followed by sonication in

529 an ice-bath. All lysates were clarified at 40,000 RCF for 45 min at 4°C and filtered. Proteins

530 were bound to HisPur Ni-NTA Resin (Thermo Scientific) and washed extensively with lysis

531 buffer with 500 mM NaCl. 6xHis-TEV-FANCI or His-3C-FANCD2 proteins were eluted by

532 lysis buffer with 250 mM imidazole and lower NaCl for ion exchange chromatography (50

533 mM NaCl). Anion exchange for FANCI and FANCD2 were all performed using a 5 ml

534 HiTRAP Q HP column (GE Life Sciences) using AKTA FPLC systems and eluted with a

535 linear gradient (50 mM Tris pH8.0, 50-1000 mM NaCl, 5 mM Dithiothreitol (DTT) and 5%

536 v/v glycerol). The proteins were further purified using SEC using a Superose6 10/300 GL

537 column in 20 mM Tris pH8.0, 400 mM NaCl, 5 mM DTT and 5% v/v glycerol. Proteins were

538 concentrated using 50,000 MWCO Amicon Ultra centrifugal filters (Merck) to before flash-

539 frozen as single-use aliquots in the same buffer system.

540 Preparation of UbIR800 material

541 6xHis-smt3-SGCGSG-Ubiquitin material post affinity and gel-filtration chromatography was

542 dialysed into 1x Phosphate buffered saline (PBS) pH7.4 with 0.5 mM EDTA. Protein

543 concentration was estimates using SDS-PAGE and SimplyBlue (Invitrogen) staining using

544 known protein standards of similar size. The dialysed material was reduced with a final

545 concentration of 50 mM 2-Mercaptoethanol (Sigma Aldrich) for 1 hour at 37°C. The material

546 was rapidly buffered exchanged into 1xPBS pH7.4 with 0.5 mM EDTA using 7K MWCO

547 Zeba Spin Desalting Columns (Pierce) and immediately mixed, at 1:2 ratio, with DyLight

548 800 Maleimide (Life Technologies) dye solubilised in neat Dimethylformamide (DMF,

549 Pierce). All subsequent steps were protected from direct light. The labelling reaction was

550 allowed to proceed at 25°C for 8-10h, quenched using 50 mM 2-Mercaptoethanol and excess

551 dye removed by extensive dialyses into 1x PBS buffer using 10,000 MWCO Spectra/Por

552 membrane (Spectrum Labs). The labelled fusion protein was cleaved using 6xHis-Ulp1

553 protease,passed over Ni-NTA Resin (Thermo Scientific) to capture the protease and 6xHis-

554 smt3 tag. The material was further purified using Superdex 75 10/300 GL in 50 mM Tris-HCl

555 pH7.5, 150 mM NaCl. Fractions with labelled ubiquitin were pooled and protein

556 concentration was estimated as above.

557 Multi-turnover substrate ubiquitination assays

558 All reactions were carried out in 50 mM Tris-HCl, 100 mM NaCl, 1 mM TCEP, 2 mM

559 MgCl2 and 4% v/v glycerol buffer system at pH7.6 and 30°C. Frozen protein aliquots were

560 thawed on ice and reactions performed within 12 h of thaw. All reaction mixtures were

561 performed on ice. The DNA co-factors for human substrates have been previously described

562 (Longerich et al., 2014) and the following sequences were synthesised as 1 μmole duplexes

563 and PAGE purified (IDT technologies).

564 Sense -

565 TTGATCAGAGGTCATTTGAATTCATGGCTTCGAGCTTCATGTAGAGTCGACGGTG

566 CTGGGATCCAACATGTTTTCAATCTG

567 Antisense -

568 CAGATTGAAAACATGTTGGATCCCAGCACCGTCGACTCTACATGAAGCTCGAAG

569 CCATGAATTCAAATGACCTCTGATCAA

570 End-point reactions (15 μL) contained 25 nM 6xHis-Ube1, 3 μM UbIR800, 0.1 μM Ube2T and

571 FANCLUR, 1 μM FANCI-FANCD2 complex or FANCI alone and 2 μM dsDNA. Reaction

572 mixes were incubated on ice for 10 min to allow for substrate-DNA complex formation

573 followed by addition of Adenosine triphosphate (ATP) at a final concentration of 5 mM.

574 Reactions were terminated after 30 min with an equal amount of 2xLDS sample buffer

575 (Pierce) containing 200 mM 2- Mercaptoethanol and boiled at 98°C for 3min. Subsequently 4

576 μL of boiled samples were resolved in 10 well Bolt 4-12% Bis-Tris Plus Gels (Invitrogen)

577 using a 1x MOPS running buffer system (Invitrogen). Gels were resolved until the 25 kDa

578 MW marker of All Blue Precision Plus protein standard (Bio-Rad) is at the bottom of the gel.

579 Gels were rinsed with water and scanned by direct fluorescence monitoring using Li-COR

580 Odyssey CLX Infrared Imaging System. Time-course reactions (60 μL) for rate

581 determination were setup as described above except the 0 min time-point was taken prior to

582 addition of ATP, 8 μL sample at indicated time-points and terminated with equal amounts of

583 2xLDS reducing sample buffer. Samples were boiled together after last time-point and 4 μL

584 were resolved in 17 well NuPAGE 4-12% Bis-Tris Gels (Invitrogen) as described above.

585 Gels were scanned as before and analysed using ImageStudio software (Li-COR). A custom

586 rectangle in 0 min lane was used for background subtraction. An identical shape area was

587 used to quantify amount of product formed using trimmed signal intensities values. The data

588 was exported into Microsoft Excel and plotted against time to determine rates in the linear

589 reaction range. Finally rates were normalized to wildtype E2. Experiments were performed in

590 triplicate and final rate graphs were plotted (Mean ± SD) in GraphPad Prism 7.

591 PCNA ubiquitination assays (20 μL) contained 25 nM 6xHis-Ube1, 20 μM Ub, 2 μM Ube2B,

592 9 μM 6xHisPCNA and the indicated amounts of 6xHis-smt3-Rad18. Post ATP addition (5

593 mM), reactions were allowed to proceed for 90 min and terminated with equal amounts of

594 2xLDS reducing sample buffer and boiled as before. Samples were diluted to 60 μL using

595 1xLDS reducing sample buffer. 3 μL (PCNA blot) and 10 μL (Ube2B and Rad18 blot) of the

596 diluted sample was resolved in 15 well NuPAGE 4-12% Bis-Tris Gels (Invitrogen) and

597 transferred using iBlot Western Blotting system and nitrocellulose transfer stacks

598 (Invitrogen). Membranes were blocked using 1xPBS buffer containing 1% w/v BSA and

599 0.05% v/v Tween20. The respective membranes were incubated overnight at 4°C with anti-

600 PCNA mouse monoclonal antibody (ab29, Abcam) at 0.4 ng/μL, anti-Ube2B rabbit

601 polyclonal (10733-1-AP, Proteintech) at 0.15 ng/μL and anti-Rad18 rabbit polyclonal

602 antibody (18333-1-AP) at 0.6 ng/μL. Blots were washed 3x15min with 1xPBS buffer with

603 0.05% v/v Tween20 and probed with IR800 labelled secondary antibodies (Li-COR) of the

604 corresponding species at 0.1 ng/μL for 2 h at room-temperature. Blots were washed 3x15min

605 with 1xPBS buffer with 0.05% v/v Tween20 and scanned by direct fluorescence monitoring

606 using Li-COR Odyssey CLX Infrared Imaging System. Experiments were performed in

607 duplicate to confirm consistency of results

608 E2 charging, autoubiquitination and lysine discharge assays

609 All reactions were carried out in 50 mM Hepes, 100 mM NaCl, 1 mM TCEP, 2 mM MgCl2

610 and 4% v/v glycerol buffer system at pH7.6 and 30°C. Ube2T charging reactions (10 μL)

611 contained 100 nM 6xHis-Ube1, 10 μM E2 and Ubiquitin. Reactions were commenced by

612 addition of buffer or ATP at a final concentration of 5 mM. Reactions were terminated after 5

613 min with 5 μL of non-reducing 3xLDS sample buffer. 1 μg of Ube2T was resolved in 15 well

614 NuPAGE 4-12% Bis-Tris Gels (Invitrogen). Gels were rinsed with water, stained with

615 InstantBlue Coomassie stain. Gels were de-stained with water and scanned. Experiments

616 were performed in duplicate to confirm consistency of results. Ube2T1-152 auto-ubiquitination

617 reactions (60 μL) contained 100 nM 6xHis-Ube1, 10 μM E2 and 20 μM Ubiquitin. The 0

618 time-point was taken prior to ATP addition (5 mM final), 8 μL sample was taken at indicated

619 time-points and terminated with 4 μL of 3xLDS reducing sample buffer. 0.8 μg of Ube2T1-152

620 was resolved in 15 well NuPAGE 12% Bis-Tris Gels (Invitrogen), stained and de-stained as

621 before. Experiments were performed in triplicate to confirm consistency of results. For lysine

622 discharge assays, charging reactions (50 μL) contained 120 nM 6xHis-Ube1, 12 μM E2 and

623 24 μM Ubiquitin and started with ATP (5 mM final). After 10 min at 30°C, 0.5 U of Apyrase

624 (NEB) was added to arrest the reaction and left at room temperature for 3 min. The chase was

625 initiated by adding 10 mM L-Lysine (Sigma Aldrich) to bring the final volume to 60 μL.

626 Initial time-point sample (8 μL) was taken at 0.1 min and subsequent samples were taken at

627 indicated time-points. Samples were terminated with 4 μL of 3xLDS non-reducing sample

628 buffer. 0.9 μg of Ube2T1-152 was resolved in 15 well NuPAGE 12% Bis-Tris Gels

629 (Invitrogen), stained and de-stained as before. Gels were scanned by direct fluorescence

630 monitoring (700 nm λ) using Li-COR Odyssey CLX Infrared Imaging System. The

631 intensities of product formed (E2 released) was obtained and converted protein amount

632 (picomoles) using an E2 only serial dilution reference gel that was stained and de-stained in

633 parallel. Absolute rates were determined by plotting product versus time in Microsoft Excel.

634 Experiments were performed in duplicate and final rate graphs were plotted (Mean ± SD) in

635 GraphPad Prism 7.

636 Single-turnover substrate ubiquitination assays

637 All reactions were carried out in 50 mM Tris-HCl, 100 mM NaCl, 1 mM TCEP, 2 mM

638 MgCl2 and 4% v/v glycerol buffer system at pH7.6. Charging reactions (30 μL) contained 10

639 nM 6xHis-Ube1, 10 μM E2 and 10 μM UbIR800 and started with ATP (5 mM final). After 10

640 min at 30°C, 0.25 U of Apyrase was added to arrest the reaction and left at room temperature

641 for 3 min. E2 charging efficiency was determined to be ~80%. Chase mixes (40 μL)

642 containing 2 μM FANCIK523R-FANCD2-dsDNA complex, 0.2 μM E2~UbIR800 and indicated

643 amounts of FANCLUR were incubated at 30°C. After 10 min, 5 μL of the reaction diluted 1-

644 fold and terminated with 10 μL of 2xLDS non-reducing or reducing sample buffer. Samples

645 with reducing agent were boiled at 98°C for 3min. 4 μL of each sample were resolved in 10

646 well Bolt 4-12% Bis-Tris Plus Gels (Invitrogen) until the 25 kDa MW marker is at the bottom

647 of the gel. Gels were rinsed with water and scanned by direct fluorescence monitoring using

648 Li-COR Odyssey CLX Infrared Imaging System.

649 Isothermal titration calorimetry

650 ITC experiments were performed using MicroCal PEAQ-ITC (Malvern instruments). All

651 experiments were performed at 20°C, in duplicate, using freshly prepared proteins within 2

652 days of the last purification step. Proteins were buffer exchanged using 7K MWCO Zeba

653 Spin Desalting Columns (Pierce) into 100 mM Tris-HCl, 100 mM NaCl, 0.4 mM TCEP

654 buffer at pH 8.0 that was filtered and thoroughly degassed. FANCLUR (ranging 22 to 34 μM)

655 and FANCLR (~32 μM) was held in the cell, while Ube2T (ranging 400 to 600 μM) was

656 present in the syringe. A total of 16 injections were carried out with the first injection of 0.3

657 μL over 0.6s followed by 15 injections of 1.5 μL over 3s. All injections were spaced by 120s

658 with mix speed set at 500 rotations per minute. Each experiment was controlled by an

659 identical E2 into buffer run to account for the heat of dilution. All data were fitted using a

660 single-site binding model using MicroCal PEAQ-ITC analysis software.

661 Residue Network Analysis

662 Residue Interaction Networks (RINs) were generated using the RIN generator webserver

663 (http://protein.bio.unipd.it/ring/)(Piovesan et al., 2016). For this study, networks were

664 generated for consecutive residues in individual chains using relaxed distance thresholds.

665 Further, all atoms of a likely residue pair were considered when applying distance thresholds

666 and subsequently all possible interactions outputted. Finally, water molecules and hetero

667 atoms/ligands were omitted for the RIN. The output files were uploaded in Cytoscape

668 (Shannon et al., 2003), and the RINs were compared using the RINalyzer app (Doncheva et

669 al., 2011) using residue IDs as matching attribute. In case of Ube2T, two copies of the RING

670 bound state were first individually compared to unbound state using a difference edge weight.

671 The resulting RIN comparison matrices were merged into a collective network using the

672 inbuilt merge network tool on Cytoscape.

673

674 Author contributions

675 V.K.C. – Conceptualization, Methodology, Validation, Formal Analysis, Investigation,

676 Resources, Writing - original draft preparation, review and editing, Visualization, Project

677 Administration.

678 C.A. – Methodology, Validation, Resources, Writing - review.

679 R.T. – Resources.

680 H.W. – Conceptualization, Writing - review and editing, Supervision, Project Administration,

681 Funding Acquisition

682

683 Acknowledgements

684 This work was supported by the Medical Research Council (MRC grant number

685 MC_UU_12016/12); the EMBO Young Investigator Programme to H.W.; the European

686 Research Council (ERC-2015-CoG-681582 ICLUb consolidator grant to H.W. The pLou3

687 Rat RNF4 RING-RING DNA construct was a kind gift from Prof. Ron Hay, University of

688 Dundee. The human PCNA cDNA template, was a kind gift from Dr Svend Petersen-Mahrt,

689 European Institute of Oncology, Milan. The Human 6xHis-Ube1 protein material was a kind

690 gift from Dr. Axel Knebel, MRC Protein Phosphorylation and Ubiquitylation Unit. All

691 constructs are available on request from the MRC Protein Phosphorylation and

692 Ubiquitylation Unit reagents Web page (http://mrcppureagents.dundee.ac.uk) or from the

693 corresponding author.

694

695 Conflict of interest

696 Authors declare no conflict of interest.

697

698 References

699 Al-Hakim, A., Escribano-Diaz, C., Landry, M.C., O'Donnell, L., Panier, S., Szilard, R.K., and Durocher, D. 700 (2010). The ubiquitous role of ubiquitin in the DNA damage response. DNA Repair (Amst) 9, 1229-1240. 701 Alpi, A.F., Pace, P.E., Babu, M.M., and Patel, K.J. (2008). Mechanistic insight into site-restricted 702 monoubiquitination of FANCD2 by Ube2t, FANCL, and FANCI. Mol Cell 32, 767-777. 703 Alpi, A.F., and Patel, K.J. (2009). Monoubiquitylation in the Fanconi anemia DNA damage response pathway. 704 DNA Repair (Amst) 8, 430-435. 705 Benirschke, R.C., Thompson, J.R., Nomine, Y., Wasielewski, E., Juranic, N., Macura, S., Hatakeyama, S., 706 Nakayama, K.I., Botuyan, M.V., and Mer, G. (2010). Molecular basis for the association of human E4B U box 707 ubiquitin ligase with E2-conjugating enzymes UbcH5c and Ubc4. Structure 18, 955-965. 708 Bentley, M.L., Corn, J.E., Dong, K.C., Phung, Q., Cheung, T.K., and Cochran, A.G. (2011). Recognition of 709 UbcH5c and the nucleosome by the Bmi1/Ring1b ubiquitin ligase complex. EMBO J 30, 3285-3297. 710 Branigan, E., Plechanovova, A., Jaffray, E.G., Naismith, J.H., and Hay, R.T. (2015). Structural basis for the 711 RING-catalyzed synthesis of K63-linked ubiquitin chains. Nat Struct Mol Biol 22, 597-602. 712 Brown, N.G., VanderLinden, R., Watson, E.R., Qiao, R., Grace, C.R., Yamaguchi, M., Weissmann, F., Frye, 713 J.J., Dube, P., Ei Cho, S., et al. (2015). RING E3 mechanism for ubiquitin ligation to a disordered substrate 714 visualized for human anaphase-promoting complex. Proc Natl Acad Sci U S A 112, 5272-5279. 715 Brown, N.G., Watson, E.R., Weissmann, F., Jarvis, M.A., VanderLinden, R., Grace, C.R.R., Frye, J.J., Qiao, R., 716 Dube, P., Petzold, G., et al. (2014). Mechanism of polyubiquitination by human anaphase-promoting complex: 717 RING repurposing for ubiquitin chain assembly. Mol Cell 56, 246-260. 718 Ceccaldi, R., Sarangi, P., and D'Andrea, A.D. (2016). The Fanconi anaemia pathway: new players and new 719 functions. Nat Rev Mol Cell Biol 17, 337-349. 720 Chakrabarti, K.S., Li, J., Das, R., and Byrd, R.A. (2017). Conformational Dynamics and Allostery in E2:E3 721 Interactions Drive Ubiquitination: gp78 and Ube2g2. Structure 25, 794-805 e795. 722 Cole, A.R., Lewis, L.P., and Walden, H. (2010). The structure of the catalytic subunit FANCL of the Fanconi 723 anemia core complex. Nat Struct Mol Biol 17, 294-298. 724 Das, R., Liang, Y.H., Mariano, J., Li, J., Huang, T., King, A., Tarasov, S.G., Weissman, A.M., Ji, X., and Byrd, 725 R.A. (2013). Allosteric regulation of E2:E3 interactions promote a processive ubiquitination machine. EMBO J 726 32, 2504-2516. 727 Das, R., Mariano, J., Tsai, Y.C., Kalathur, R.C., Kostova, Z., Li, J., Tarasov, S.G., McFeeters, R.L., Altieri, 728 A.S., Ji, X., et al. (2009). Allosteric activation of E2-RING finger-mediated ubiquitylation by a structurally 729 defined specific E2-binding region of gp78. Mol Cell 34, 674-685. 730 Doncheva, N.T., Klein, K., Domingues, F.S., and Albrecht, M. (2011). Analyzing and visualizing residue 731 networks of protein structures. Trends Biochem Sci 36, 179-182. 732 Dou, H., Buetow, L., Sibbet, G.J., Cameron, K., and Huang, D.T. (2012). BIRC7-E2 ubiquitin conjugate 733 structure reveals the mechanism of ubiquitin transfer by a RING dimer. Nat Struct Mol Biol 19, 876-883. 734 Eddins, M.J., Carlile, C.M., Gomez, K.M., Pickart, C.M., and Wolberger, C. (2006). Mms2-Ubc13 covalently 735 bound to ubiquitin reveals the structural basis of linkage-specific polyubiquitin chain formation. Nat Struct Mol 736 Biol 13, 915-920. 737 Ernst, A., Avvakumov, G., Tong, J., Fan, Y., Zhao, Y., Alberts, P., Persaud, A., Walker, J.R., Neculai, A.M., 738 Neculai, D., et al. (2013). A strategy for modulation of enzymes in the ubiquitin system. Science 339, 590-595. 739 Freemont, P.S. (2000). RING for destruction? Curr Biol 10, R84-87. 740 Gabrielsen, M., Buetow, L., Nakasone, M.A., Ahmed, S.F., Sibbet, G.J., Smith, B.O., Zhang, W., Sidhu, S.S., 741 and Huang, D.T. (2017). A General Strategy for Discovery of Inhibitors and Activators of RING and U-box E3 742 Ligases with Ubiquitin Variants. Mol Cell 68, 456-470 e410. 743 Gallego, L.D., Ghodgaonkar Steger, M., Polyansky, A.A., Schubert, T., Zagrovic, B., Zheng, N., Clausen, T., 744 Herzog, F., and Kohler, A. (2016). Structural mechanism for the recognition and ubiquitination of a single 745 nucleosome residue by Rad6-Bre1. Proc Natl Acad Sci U S A 113, 10553-10558. 746 Garaycoechea, J.I., and Patel, K.J. (2014). Why does the bone marrow fail in Fanconi anemia? Blood 123, 26- 747 34. 748 Garcia-Higuera, I., Taniguchi, T., Ganesan, S., Meyn, M.S., Timmers, C., Hejna, J., Grompe, M., and D'Andrea, 749 A.D. (2001). Interaction of the Fanconi anemia proteins and BRCA1 in a common pathway. Mol Cell 7, 249- 750 262.

751 Hewitt, W.M., Lountos, G.T., Zlotkowski, K., Dahlhauser, S.D., Saunders, L.B., Needle, D., Tropea, J.E., Zhan, 752 C., Wei, G., Ma, B., et al. (2016). Insights Into the Allosteric Inhibition of the SUMO E2 Enzyme Ubc9. Angew 753 Chem Int Ed Engl 55, 5703-5707. 754 Hibbert, R.G., Huang, A., Boelens, R., and Sixma, T.K. (2011). E3 ligase Rad18 promotes monoubiquitination 755 rather than ubiquitin chain formation by E2 enzyme Rad6. Proc Natl Acad Sci U S A 108, 5590-5595. 756 Hira, A., Yoshida, K., Sato, K., Okuno, Y., Shiraishi, Y., Chiba, K., Tanaka, H., Miyano, S., Shimamoto, A., 757 Tahara, H., et al. (2015). Mutations in the gene encoding the E2 conjugating enzyme UBE2T cause Fanconi 758 anemia. Am J Hum Genet 96, 1001-1007. 759 Hochstrasser, M. (2009). Origin and function of ubiquitin-like proteins. Nature 458, 422-429. 760 Hodson, C., Cole, A.R., Lewis, L.P., Miles, J.A., Purkiss, A., and Walden, H. (2011). Structural analysis of 761 human FANCL, the E3 ligase in the Fanconi anemia pathway. J Biol Chem 286, 32628-32637. 762 Hodson, C., Purkiss, A., Miles, J.A., and Walden, H. (2014). Structure of the human FANCL RING-Ube2T 763 complex reveals determinants of cognate E3-E2 selection. Structure 22, 337-344. 764 Huang, A., Hibbert, R.G., de Jong, R.N., Das, D., Sixma, T.K., and Boelens, R. (2011). Symmetry and 765 asymmetry of the RING-RING dimer of Rad18. J Mol Biol 410, 424-435. 766 Joo, W., Xu, G., Persky, N.S., Smogorzewska, A., Rudge, D.G., Buzovetsky, O., Elledge, S.J., and Pavletich, 767 N.P. (2011). Structure of the FANCI-FANCD2 complex: insights into the Fanconi anemia DNA repair pathway. 768 Science 333, 312-316. 769 Kelly, A., Wickliffe, K.E., Song, L., Fedrigo, I., and Rape, M. (2014). Ubiquitin chain elongation requires E3- 770 dependent tracking of the emerging conjugate. Mol Cell 56, 232-245. 771 Kim, W., Bennett, E.J., Huttlin, E.L., Guo, A., Li, J., Possemato, A., Sowa, M.E., Rad, R., Rush, J., Comb, M.J., 772 et al. (2011). Systematic and quantitative assessment of the ubiquitin-modified proteome. Mol Cell 44, 325-340. 773 Kottemann, M.C., and Smogorzewska, A. (2013). Fanconi anaemia and the repair of Watson and Crick DNA 774 crosslinks. Nature 493, 356-363. 775 Kulathu, Y., and Komander, D. (2012). Atypical ubiquitylation - the unexplored world of polyubiquitin beyond 776 Lys48 and Lys63 linkages. Nat Rev Mol Cell Biol 13, 508-523. 777 Li, S., Liang, Y.H., Mariano, J., Metzger, M.B., Stringer, D.K., Hristova, V.A., Li, J., Randazzo, P.A., Tsai, 778 Y.C., Ji, X., et al. (2015). Insights into Ubiquitination from the Unique Clamp-like Binding of the RING E3 779 AO7 to the E2 UbcH5B. J Biol Chem 290, 30225-30239. 780 Li, W., Tu, D., Li, L., Wollert, T., Ghirlando, R., Brunger, A.T., and Ye, Y. (2009). Mechanistic insights into 781 active site-associated polyubiquitination by the ubiquitin-conjugating enzyme Ube2g2. Proc Natl Acad Sci U S 782 A 106, 3722-3727. 783 Liang, C.C., Li, Z., Lopez-Martinez, D., Nicholson, W.V., Venien-Bryan, C., and Cohn, M.A. (2016). The 784 FANCD2-FANCI complex is recruited to DNA interstrand crosslinks before monoubiquitination of FANCD2. 785 Nat Commun 7, 12124. 786 Liu, W., Shang, Y., Zeng, Y., Liu, C., Li, Y., Zhai, L., Wang, P., Lou, J., Xu, P., Ye, Y., et al. (2014). Dimeric 787 Ube2g2 simultaneously engages donor and acceptor ubiquitins to form Lys48-linked ubiquitin chains. EMBO J 788 33, 46-61. 789 Longerich, S., Kwon, Y., Tsai, M.S., Hlaing, A.S., Kupfer, G.M., and Sung, P. (2014). Regulation of FANCD2 790 and FANCI monoubiquitination by their interaction and by DNA. Nucleic Acids Res 42, 5657-5670. 791 Machida, Y.J., Machida, Y., Chen, Y., Gurtan, A.M., Kupfer, G.M., D'Andrea, A.D., and Dutta, A. (2006). 792 UBE2T is the E2 in the Fanconi anemia pathway and undergoes negative autoregulation. Mol Cell 23, 589-596. 793 Mattiroli, F., Uckelmann, M., Sahtoe, D.D., van Dijk, W.J., and Sixma, T.K. (2014). The nucleosome acidic 794 patch plays a critical role in RNF168-dependent ubiquitination of histone H2A. Nat Commun 5, 3291. 795 Mattiroli, F., Vissers, J.H., van Dijk, W.J., Ikpa, P., Citterio, E., Vermeulen, W., Marteijn, J.A., and Sixma, T.K. 796 (2012). RNF168 ubiquitinates K13-15 on H2A/H2AX to drive DNA damage signaling. Cell 150, 1182-1195. 797 McGinty, R.K., Henrici, R.C., and Tan, S. (2014). Crystal structure of the PRC1 ubiquitylation module bound to 798 the nucleosome. Nature 514, 591-596. 799 Meetei, A.R., de Winter, J.P., Medhurst, A.L., Wallisch, M., Waisfisz, Q., van de Vrugt, H.J., Oostra, A.B., 800 Yan, Z., Ling, C., Bishop, C.E., et al. (2003). A novel ubiquitin ligase is deficient in Fanconi anemia. Nat Genet 801 35, 165-170. 802 Metzger, M.B., Liang, Y.H., Das, R., Mariano, J., Li, S., Li, J., Kostova, Z., Byrd, R.A., Ji, X., and Weissman, 803 A.M. (2013). A structurally unique E2-binding domain activates ubiquitination by the ERAD E2, Ubc7p, 804 through multiple mechanisms. Mol Cell 50, 516-527. 805 Metzger, M.B., Pruneda, J.N., Klevit, R.E., and Weissman, A.M. (2014). RING-type E3 ligases: master 806 manipulators of E2 ubiquitin-conjugating enzymes and ubiquitination. Biochim Biophys Acta 1843, 47-60. 807 Middleton, A.J., and Day, C.L. (2015). The molecular basis of lysine 48 ubiquitin chain synthesis by Ube2K. 808 Sci Rep 5, 16793.

809 Morreale, F.E., Bortoluzzi, A., Chaugule, V.K., Arkinson, C., Walden, H., and Ciulli, A. (2017). Allosteric 810 Targeting of the Fanconi Anemia Ubiquitin-Conjugating Enzyme Ube2T by Fragment Screening. J Med Chem 811 60, 4093-4098. 812 Nijman, S.M., Huang, T.T., Dirac, A.M., Brummelkamp, T.R., Kerkhoven, R.M., D'Andrea, A.D., and 813 Bernards, R. (2005). The deubiquitinating enzyme USP1 regulates the Fanconi anemia pathway. Mol Cell 17, 814 331-339. 815 Ozkan, E., Yu, H., and Deisenhofer, J. (2005). Mechanistic insight into the allosteric activation of a ubiquitin- 816 conjugating enzyme by RING-type ubiquitin ligases. Proc Natl Acad Sci U S A 102, 18890-18895. 817 Petroski, M.D., and Deshaies, R.J. (2005). Mechanism of lysine 48-linked ubiquitin-chain synthesis by the 818 cullin-RING ubiquitin-ligase complex SCF-Cdc34. Cell 123, 1107-1120. 819 Pickart, C.M. (2001). Mechanisms underlying ubiquitination. Annu Rev Biochem 70, 503-533. 820 Piovesan, D., Minervini, G., and Tosatto, S.C. (2016). The RING 2.0 web server for high quality residue 821 interaction networks. Nucleic Acids Res 44, W367-374. 822 Plechanovova, A., Jaffray, E.G., McMahon, S.A., Johnson, K.A., Navratilova, I., Naismith, J.H., and Hay, R.T. 823 (2011). Mechanism of ubiquitylation by dimeric RING ligase RNF4. Nat Struct Mol Biol 18, 1052-1059. 824 Plechanovova, A., Jaffray, E.G., Tatham, M.H., Naismith, J.H., and Hay, R.T. (2012). Structure of a RING E3 825 ligase and ubiquitin-loaded E2 primed for catalysis. Nature 489, 115-120. 826 Pruneda, J.N., Littlefield, P.J., Soss, S.E., Nordquist, K.A., Chazin, W.J., Brzovic, P.S., and Klevit, R.E. (2012). 827 Structure of an E3:E2~Ub complex reveals an allosteric mechanism shared among RING/U-box ligases. Mol 828 Cell 47, 933-942. 829 Rajendra, E., Oestergaard, V.H., Langevin, F., Wang, M., Dornan, G.L., Patel, K.J., and Passmore, L.A. (2014). 830 The genetic and biochemical basis of FANCD2 monoubiquitination. Mol Cell 54, 858-869. 831 Rickman, K.A., Lach, F.P., Abhyankar, A., Donovan, F.X., Sanborn, E.M., Kennedy, J.A., Sougnez, C., 832 Gabriel, S.B., Elemento, O., Chandrasekharappa, S.C., et al. (2015). Deficiency of UBE2T, the E2 Ubiquitin 833 Ligase Necessary for FANCD2 and FANCI Ubiquitination, Causes FA-T Subtype of Fanconi Anemia. Cell Rep 834 12, 35-41. 835 Rodrigo-Brenni, M.C., Foster, S.A., and Morgan, D.O. (2010). Catalysis of lysine 48-specific ubiquitin chain 836 assembly by residues in E2 and ubiquitin. Mol Cell 39, 548-559. 837 Rodrigo-Brenni, M.C., and Morgan, D.O. (2007). Sequential E2s drive polyubiquitin chain assembly on APC 838 targets. Cell 130, 127-139. 839 Saha, A., Lewis, S., Kleiger, G., Kuhlman, B., and Deshaies, R.J. (2011). Essential role for ubiquitin-ubiquitin- 840 conjugating enzyme interaction in ubiquitin discharge from Cdc34 to substrate. Mol Cell 42, 75-83. 841 Sareen, A., Chaudhury, I., Adams, N., and Sobeck, A. (2012). Fanconi anemia proteins FANCD2 and FANCI 842 exhibit different DNA damage responses during S-phase. Nucleic Acids Res 40, 8425-8439. 843 Sato, K., Toda, K., Ishiai, M., Takata, M., and Kurumizaka, H. (2012). DNA robustly stimulates FANCD2 844 monoubiquitylation in the complex with FANCI. Nucleic Acids Res 40, 4553-4561. 845 Shannon, P., Markiel, A., Ozier, O., Baliga, N.S., Wang, J.T., Ramage, D., Amin, N., Schwikowski, B., and 846 Ideker, T. (2003). Cytoscape: a software environment for integrated models of biomolecular interaction 847 networks. Genome Res 13, 2498-2504. 848 Sheng, Y., Hong, J.H., Doherty, R., Srikumar, T., Shloush, J., Avvakumov, G.V., Walker, J.R., Xue, S., 849 Neculai, D., Wan, J.W., et al. (2012). A human ubiquitin conjugating enzyme (E2)-HECT E3 ligase structure- 850 function screen. Mol Cell Proteomics 11, 329-341. 851 Sims, A.E., Spiteri, E., Sims, R.J., 3rd, Arita, A.G., Lach, F.P., Landers, T., Wurm, M., Freund, M., Neveling, 852 K., Hanenberg, H., et al. (2007). FANCI is a second monoubiquitinated member of the Fanconi anemia 853 pathway. Nat Struct Mol Biol 14, 564-567. 854 Smogorzewska, A., Matsuoka, S., Vinciguerra, P., McDonald, E.R., 3rd, Hurov, K.E., Luo, J., Ballif, B.A., 855 Gygi, S.P., Hofmann, K., D'Andrea, A.D., et al. (2007). Identification of the FANCI protein, a 856 monoubiquitinated FANCD2 paralog required for DNA repair. Cell 129, 289-301. 857 Stewart, M.D., Ritterhoff, T., Klevit, R.E., and Brzovic, P.S. (2016). E2 enzymes: more than just middle men. 858 Cell Res 26, 423-440. 859 Sugahara, R., Mon, H., Lee, J.M., and Kusakabe, T. (2012). Monoubiquitination-dependent chromatin loading 860 of FancD2 in silkworms, a species lacking the FA core complex. Gene 501, 180-187. 861 Swuec, P., Renault, L., Borg, A., Shah, F., Murphy, V.J., van Twest, S., Snijders, A.P., Deans, A.J., and Costa, 862 A. (2017). The FA Core Complex Contains a Homo-dimeric Catalytic Module for the Symmetric Mono- 863 ubiquitination of FANCI-FANCD2. Cell Rep 18, 611-623. 864 Turco, E., Gallego, L.D., Schneider, M., and Kohler, A. (2015). Monoubiquitination of histone H2B is intrinsic 865 to the Bre1 RING domain-Rad6 interaction and augmented by a second Rad6-binding site on Bre1. J Biol Chem 866 290, 5298-5310. 867 Uckelmann, M., and Sixma, T.K. (2017). Histone ubiquitination in the DNA damage response. DNA Repair 868 (Amst) 56, 92-101.

869 Udeshi, N.D., Svinkina, T., Mertins, P., Kuhn, E., Mani, D.R., Qiao, J.W., and Carr, S.A. (2013). Refined 870 preparation and use of anti-diglycine remnant (K-epsilon-GG) antibody enables routine quantification of 871 10,000s of ubiquitination sites in single proteomics experiments. Mol Cell Proteomics 12, 825-831. 872 Ulrich, H.D., and Walden, H. (2010). Ubiquitin signalling in DNA replication and repair. Nat Rev Mol Cell Biol 873 11, 479-489. 874 Valimberti, I., Tiberti, M., Lambrughi, M., Sarcevic, B., and Papaleo, E. (2015). E2 superfamily of ubiquitin- 875 conjugating enzymes: constitutively active or activated through phosphorylation in the catalytic cleft. Sci Rep 5, 876 14849. 877 van den Ent, F., and Lowe, J. (2006). RF cloning: a restriction-free method for inserting target genes into 878 plasmids. J Biochem Biophys Methods 67, 67-74. 879 van Twest, S., Murphy, V.J., Hodson, C., Tan, W., Swuec, P., O'Rourke, J.J., Heierhorst, J., Crismani, W., and 880 Deans, A.J. (2017). Mechanism of Ubiquitination and Deubiquitination in the Fanconi Anemia Pathway. Mol 881 Cell 65, 247-259. 882 Virts, E.L., Jankowska, A., Mackay, C., Glaas, M.F., Wiek, C., Kelich, S.L., Lottmann, N., Kennedy, F.M., 883 Marchal, C., Lehnert, E., et al. (2015). AluY-mediated germline deletion, duplication and somatic stem cell 884 reversion in UBE2T defines a new subtype of Fanconi anemia. Hum Mol Genet 24, 5093-5108. 885 Walden, H., and Deans, A.J. (2014). The Fanconi anemia DNA repair pathway: structural and functional 886 insights into a complex disorder. Annu Rev Biophys 43, 257-278. 887 Wickliffe, K.E., Lorenz, S., Wemmer, D.E., Kuriyan, J., and Rape, M. (2011). The mechanism of linkage- 888 specific ubiquitin chain elongation by a single-subunit E2. Cell 144, 769-781. 889 Windheim, M., Peggie, M., and Cohen, P. (2008). Two different classes of E2 ubiquitin-conjugating enzymes 890 are required for the mono-ubiquitination of proteins and elongation by polyubiquitin chains with a specific 891 topology. Biochem J 409, 723-729. 892 Yunus, A.A., and Lima, C.D. (2006). Lysine activation and functional analysis of E2-mediated conjugation in 893 the SUMO pathway. Nat Struct Mol Biol 13, 491-499. 894 Zhang, W., Wu, K.P., Sartori, M.A., Kamadurai, H.B., Ordureau, A., Jiang, C., Mercredi, P.Y., Murchie, R., 895 Hu, J., Persaud, A., et al. (2016). System-Wide Modulation of HECT E3 Ligases with Selective Ubiquitin 896 Variant Probes. Mol Cell 62, 121-136. 897 Zhang, X.Y., Langenick, J., Traynor, D., Babu, M.M., Kay, R.R., and Patel, K.J. (2009). Xpf and not the 898 Fanconi anaemia proteins or Rev3 accounts for the extreme resistance to cisplatin in Dictyostelium discoideum. 899 PLoS Genet 5, e1000645. 900

Table 1. Thermodynamic properties FANCL - Ube2T interaction1

∆H ∆G -T∆S E3 E2 Kd (nM) N (kcal.mol-1) (kcal.mol-1) (kcal.mol-1)

FANCLUR Wildtype 116.5 ± 2.6 0.98 6.46 ± 0.07 - 9.30 -15.75 ± 0.05

D32A, D33A, L92A 121.5 ± 2.9 0.94 -1.65 ± 0.16 - 9.28 - 7.64 ± 0.17

T23R, W25Q 200.0 ± 2.7 1.00 3.90 ± 0.70 - 8.99 -12.55 ± 0.15

E54R 119.5 ± 2.2 0.93 5.60 ± 0.50 - 9.29 -14.90 ± 0.10

FANCLR Wildtype 249.5 ± 7.5 1.05 3.83 ± 0.36 - 8.86 - 12.70 ± 0.40

D32A, D33A, L92A 251.5 ± 1.5 0.95 - 2.88 ± 0.02 - 8.86 - 5.98 ± 0.03

T23R, W25Q 259.0 ± 8.4 1.05 3.89 ± 0.05 - 8.84 - 12.75 ± 0.05

901 1Average values from two experiments

902 Supplementary Table 1. Edge profile of Ube2T allosteric network model depicted in Figure

903 4a.

904 Supplementary Dataset S1. Merged Ube2T intra-residue network.

905 Supplementary Dataset S2. Merged Ube2B intra-residue network.

906 907 Figure Legends 908 Figure 1. FANCLUR mediated FANCD2 ubiquitination does not require the ubiquitin

909 Ile44-patch.

910 A Time-course multi-turnover ubiquitination assays with fluorescently labelled ubiquitin

911 (UbIR800) showing FANCLUR (0.1 μM) and Ube2T (0.1 μM) mediated site-specific

912 mono-ubiquitination of Lys561 FANCD2 (1.0 μM) when present as a FANCD2-

913 FANCI-dsDNA (1:1:2 μM) complex.

914 B Titration of FANCLUR - Ube2T enzymes on FANCD2-FANCI-dsDNA (1:1:2 μM)

915 complexes with single target lysine i.e. Lys561 FANCD2 or Lys523 FANCI reveals

916 FANCI site can be targeted but FANCD2 site is preferred.

917 C Time-course multi-turnover ubiquitination assays with fluorescently labelled ubiquitin

918 (UbIR800) showing FANCLUR (0.1 μM) and Ube2T (0.1 μM) mediated site-specific

919 mono-ubiquitination of Lys523 FANCI (1.0 μM) when present as a FANCI-dsDNA

920 (1:2 μM) complex.

921 D Comparing FANCD2-FANCI-dsDNA (1:1:2 μM) ubiquitination activities of the

922 FANCLUR (0.1 μM) - Ube2T (0.1 μM) pair with RNF4RING-RING (0.1 μM) - Ube2D3

923 (0.2 μM) pair using fluorescently labelled wildtype and Ile44Ala ubiquitin. RNF4RING-

924 RING and Ube2D3 robustly ubiquitinates FANCI-FANCD2 using wiltype ubiquitin,

925 but is dramatically impaired by the Ile44Ala mutant while the FANCLUR and Ube2T

926 maintain activity and site-specificity even with the ubiquitin mutant. Substrate

927 ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX).

928 Figure 2. FANCL induced dynamics of Ube2T loop2 and loop7 is required for substrate

929 ubiquitination.

930 A Superpose of FANCLR (teal surface) bound copies of Ube2T (olive ribbon, PDB ID

931 4ccg.A and brown ribbon, PDB ID 4ccg.B) with the unbound Ube2T (grey ribbon)

932 structure (PDB ID 1yh2) showing little overall structural change. Close-up of helix1-

933 loop2 region (top) and loop7-loop8 region (bottom) reveal local changes. Molecular

934 figures prepared in PyMOL (Schrödinger, LLC).

935 B End-point (30 min) multi-turnover ubiquitination assay with fluorescently labelled

936 ubiquitin (UbIR800) show conserved residues in Ube2T helix1, loop2 and loop7 are

937 required for FANCLUR mediated FANCD2 and FANCI ubiquitination. Substrate

938 ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX).

939 C Effect of Ube2T helix1, loop2 and loop7 mutants on rates of FANCD2 and FANCI

940 ubiquitination. A loop2-loop7 hybrid mutant (Asp32Ala, Asp33Ala, Leu92Ala)

941 shows 75% loss in substrate ubiquitination rates. Rates normalized to wildtype levels

942 and plotted as mean ± SD (n=3).

943 D FANCL independent Ube2T1-152 autoubiquitination assay shows no effect of the

944 loop2-loop7 hybrid mutant in Lys91 autoubiquitination.

945 E Single-turnover ubiquitination assay (10 min) of a FANCD2-FANCIK523R-dsDNA

946 (2:2:2 μM) complex with increasing amounts of FANCLUR and 200 nM of Ube2T1-152,

947 K91R ~ Ub IR800 thioester or Ube2T1-152, K91R loop2-loop7 hybrid mutant ~ Ub IR800

948 thioester shows the latter is defective in modifying Lys561 FANCD2.

949 F Thermodynamics of FANCLUR interaction with Ube2T wildtype (left) and Ube2T

950 loop2-loop7 hybrid mutant (middle) shows no change in binding affinity but

951 divergent binding enthalpy. The cost of favourable enthalpy for the mutant is offset by

952 reduced conformational entropy thus contributing to reduced activity. Graphs (right)

953 plotted as mean ± range (n=2).

954 Figure 3. Novel role of Ube2T backside in FANCL mediated substrate ubiquitination.

955 A Superpose of FANCLR (teal surface) bound copy of Ube2T (olive ribbon, PDB ID

956 4ccg.A) with unbound Ube2T (grey ribbon) structure (PDB ID 1yh2) showing

957 residues on E2 backside (β1 and β2) that are repositioned upon FANCL binding.

958 Molecular figures prepared in PyMOL (Schrödinger, LLC).

959 B Ube2T backside β1 double mutant (Thr23Arg, Trp25Gln) reveals binding defect with

960 FANCLUR but is unaffected in FANCLR interaction. This suggests Ube2T backside

961 supports additional FANCL interactions.

962 C End-point (30 min) multi-turnover ubiquitination assay with fluorescently labelled

963 ubiquitin (UbIR800) show Ube2T backside mutants are more defected in FANCLUR

964 mediated FANCD2 ubiquitination than FANCI ubiquitination. Substrate

965 ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX).

966 D Effect of Ube2T backside mutants on the rates of FANCD2 and FANCI

967 ubiquitination. Backside mutants reduce FANCI modification rate to levels observed

968 in the no E3 setup. In contrast, a Ube2T backside β1 double mutant (Thr23Arg,

969 Trp25Gln) shows 75% loss in FANCD2 ubiquitination. Rates normalized to wildtype

970 levels and plotted as mean ± SD (n=3).

971 E FANCL independent Ube2T1-152 autoubiquitination assay shows no effect of the

972 backside β1 double mutant in Lys91 autoubiquitination.

973 F Single-turnover ubiquitination assay (10 min) of a FANCD2-FANCIK523R-dsDNA

974 (2:2:2 μM) complex with increasing amounts of FANCLUR and 200 nM of Ube2T1-152,

975 K91R ~ Ub IR800 thioester or Ube2T1-152, K91R β1 mutant ~ Ub IR800 thioester shows the

976 defect in the latter cannot be completely rescued by increasing FANCLUR levels.

977 Figure 4. Allosteric residue network reveals Ube2T active-site residues critical for

978 substrate ubiquitination.

979 A Allosteric network model shows dynamic rewiring of Ube2T intra-residue

980 connections upon FANCLR binding. Dashed and orange lines depict Ube2T edges in

981 unbound and FANCLR bound state respectively. Grey nodes have relative solvent

982 accessibility of less than 10% in unbound Ube2T. Nodes involved in RING binding

983 are clustered in a cyan box while those predicted to support backside interaction are

984 clustered in a purple box. Nodes in loop7 and loop8 are in unshaded boxes. Red node

985 denotes the catalytic cysteine (Cys86) while the network termini nodes (Arg84,

986 Lys91, Lys95, Asp122 and Leu124) that are within 10 Å of Cys86 have a thick

987 outline. Also depicted is the catalytic beta-element.

988 B End-point (30 min) multi-turnover ubiquitination assay with fluorescently labelled

989 ubiquitin (UbIR800) show mutations of the Ube2T network termini residues are

990 detrimental to FANCLUR mediated FANCD2 and FANCI ubiquitination. Substrate

991 ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey

992 CLX).

993 C Ubiquitin charging assays of the network termini mutants show the Leu124Ala

994 mutant alone is defected E1-based E2~Ub thioester formation.

995 D Lysine discharge assays show network termini mutants in loop7 (Lys91Ala +

996 Lys95Ala) and loop8 (Asp122Ala) have catalytic defects while Arg84Ala is not

997 defected. Graphs depict mean ± SD (n=2).

998 E End-point (30 min) multi-turnover ubiquitination assay with fluorescently labelled

999 ubiquitin (UbIR800) show the requirement of Arg/Lys residues in loop7 while Arg84,

1000 in the catalytic β-element, is critical for FANCI and FANCD2 ubiquitination. Partial

1001 compensation of activity for Ube2T Arg84Lys occurs only when the loop7 bears

1002 longer Arg residues.

1003 Figure 5. Ube2T deregulation leads to enhanced FANCL driven substrate

1004 ubiquitination.

1005 A Comparison of a network scheme (left) with structures of unbound (middle) and RING

1006 bound (right) Ube2T depicting the proposed allosteric conduit. A gating role is

1007 proposed for Glu54 (thick outline) for its regulation of Arg69. An effector role is

1008 proposed for Arg69 for its role in stabilising the catalytic β-element leading to a

1009 release of Arg84. Dashed and orange lines depict Ube2T edges in unbound and

1010 FANCLR bound state respectively. Grey nodes have relative solvent accessibility of

1011 less than 10% in unbound Ube2T. Molecular figures prepared in PyMOL

1012 (Schrödinger, LLC).

1013 B End-point (30 min) multi-turnover ubiquitination assay with fluorescently labelled

1014 ubiquitin (UbIR800) shows Glu54 gating role is dependent on its negative side chain.

1015 Removal of this charge leads to improved FANCLUR mediated substrate

1016 ubiquitination. Substrate ubiquitination is analysed by direct fluorescence monitoring

1017 (Li-COR Odyssey CLX).

1018 C End-point (30 min) multi-turnover ubiquitination assay with fluorescently labelled

1019 ubiquitin (UbIR800) shows effector role for Arg69 is linked to its positive side chain and

1020 is regulated by the Glu54 gate. While removal of the Glu54 negative charge improves

1021 improved FANCLUR mediated substrate ubiquitination, the removal of the Arg69

1022 positive charge counters this effect.

1023 D End-point (30 min) multi-turnover ubiquitination assay of a FANCD2-FANCIK523R-

1024 dsDNA (1:1:2 μM) complex with fluorescently labelled ubiquitin (UbIR800) shows a

1025 permissive gate at position 54 (Glu54Arg) can rescue FANCD2 ubiquitination defects

1026 arising from the loop2-loop7 hybrid mutant (Asp32Ala, Asp33Ala, Leu92Ala), the

1027 backside β1 double mutant (Thr23Arg, Trp25Gln) as well as the RING binding mutant

1028 (Phe63Ala). Despite deregulation of the RING binding Ube2T Glu54Arg + Phe63Ala

1029 double mutant, FANCL is still required for efficient FANCD2 ubiquitination.

1030 E Effects of the Ube2Tv1 variant with a permissive gate (Glu54Arg) alongside the

1031 Ube2Tv3 variant containing a permissive gate and flexible loop7 (Glu54Arg,

1032 Pro93Gly, Pro94Gly) on the rates of FANCLUR mediated FANCD2 (FANCIK523R-

1033 FANCD2-dsDNA complex) and FANCI (FANCI-dsDNA complex) ubiquitination.

1034 F End-point (30 min) multi-turnover ubiquitination assay of isolated FANCD2 and

1035 FANCI substrates with no DNA cofactors. Titration of FANCLUR - Ube2Tv3 enzymes

1036 shows how substrate ubiquitination is enhanced by the deregulated E2 without

1037 compromising on site-specificity.

1038 G Allosteric model for FANCLUR mediated activation of Ube2T. FANCL binding at

1039 classical RING-E2 interface and the backside E2 interface triggers long-range and

1040 short-range rewiring of Ube2T residue networks that culminate in substrate

1041 ubiquitination. The dynamic allosteric network is intrinsically regulated by conserved

1042 Ube2T elements, the allosteric β3 gating residue and a rigid loop7.

1043 Supplementary Figure 1. Solution and sequence profile of FANCLUR with Ube2T

1044 sequence conservation.

1045 A Analytical size-exclusion chromatography of FANCLUR using Superdex75 10/300 GL

1046 column (blue trace). Gel-filtration standards (dashed trace) with known elution profile

1047 were also run in same buffer system as FANCLUR and the traces were overlaid. The

1048 FANCLUR fragment resolves as a monomeric protein however, the moderate shift to

1049 an earlier elution volume suggests the central UBC-RWD and C-terminal RING

1050 domain adopt an extended conformation as observed in the fly FANCL structure

1051 (PDB ID 3k1l).

1052 B Sequencing alignment of the C-terminal regions of various RING domains that have

1053 been characterised to have a linchpin Arg residue (blue highlight). FANCL appears to

1054 lack such a residue at the analogous position. Structure of the RNF4 linchpin residue

1055 (Arg181, green cartoon, PDB ID 4ap4) shows how the linchpin contacts both the E2

1056 (Ube2D1, orange cartoon) and donor ubiquitin (yellow cartoon). In contrast, the

1057 equivalent FANCL residue Ser363 would not stabilise such conformation. Zinc co-

1058 ordinating cysteine residues are in bold. PDB IDs of respective structures are listed

1059 alongside. Molecular figures prepared in PyMOL (Schrödinger, LLC).

1060 C Sequence conservation of the UBC fold between Ube2T homologs. Residues shaded

1061 in red to yellow to highlight conservation, where red corresponds to strict

1062 conservation. Depicted above the sequences (grey) are secondary structure elements

1063 of human Ube2T are based on PDB ID 1yh2. Red star indicates catalytic cysteine

1064 while coloured circles indicate human Ube2T residues mutated in this study; blue in

1065 Figure 2, cyan in Figure 3, purple in Figure 4 and green in Figure 5.

1066 Supplementary Figure 2. Activity and binding profile of Ube2T loop2 and loop7

1067 residues.

1068 A Representative time-course ubiquitination assays with fluorescently labelled ubiquitin

1069 (UbIR800) used to derive substrate ubiquitination rates depicted in Figure 2C. Substrate

1070 ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX).

1071 B Ubiquitin charging assays of Ube2T loop2 and loop7 mutants show that the mutants

1072 do not affect E1-based E2~Ub thioester formation.

1073 C Thermodynamics of FANCLR interaction with Ube2T wildtype (left) and Ube2T

1074 loop2-loop7 hybrid mutant (middle) shows similar contrast in binding enthalpy profile

1075 as observed with FANCLUR. Graphs (right) plotted as mean ± range (n=2).

1076 Supplementary Figure 3. Activity profile of Ube2T backside residues.

1077 A Representative time-course ubiquitination assays with fluorescently labelled ubiquitin

1078 (UbIR800) used to derive substrate ubiquitination rates depicted in Figure 3D. Substrate

1079 ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX).

1080 B Ubiquitin charging assays of Ube2T backside mutants show that the mutants do not

1081 affect E1-based E2~Ub thioester formation.

1082 Supplementary Figure 4. Ube2T residue network profile and characterization of

1083 network termini residues

1084 A Summary of nodes and edges (top left) in the individual Ube2T residue interaction

1085 networks (RINs). Mosaic plots of edge distribution profile derived from RIN

1086 comparison (bottom left). Complete profile of edges and nodes obtained after merging

1087 the RIN comparison networks. Dashed and orange lines depict Ube2T edges in

1088 unbound and FANCLR bound state respectively. Thin grey lines depict edges that are

1089 shared in both networks. Grey nodes have relative solvent accessibility of less than

1090 10% in unbound Ube2T. Red node denotes the catalytic cysteine (Cys86).

1091 B Structural overlay (top) of unbound and RING bound Ube2T structures focussing on

1092 network termini residues surrounding the catalytic cysteine. Sequence alignment

1093 (bottom) of Ube2T catalytic site with various other human E2s reveal how the termini

1094 of Ube2T network (indicated by boxes) vary among the enzyme family. Catalytic

1095 cysteine is shown in red. Molecular figures prepared in PyMOL (Schrödinger, LLC).

1096 C Representative gels of lysine discharge assays plotted in Figure 4D. Coomassie

1097 stained gels are visualized by direct fluorescence monitoring in the red channel (Li-

1098 COR Odyssey CLX).

1099 D Total network profile of Arg84 residue present in Ube2T catalytic β-element. Listed

1100 alongside are edges that involve the Arg84 side chain. Edge depiction is same as in

1101 panel A.

1102 E Acidic/polar residues found proximal to target lysine on FANCD2 (left, green

1103 cartoon) and FANCI (right, blue cartoon) based on the mouse FANCI-FANCD2

1104 complex structure (PDB ID 3s4w). Sequence alignment (middle) shows conservation

1105 of these acidic/polar residues (boxed). Respective target lysine is depicted in blue.

1106 Molecular figures prepared in PyMOL (Schrödinger, LLC).

1107 Supplementary Figure 5. Activity profile of deregulated Ube2B and Ube2T variants

1108 A Comparison of a network scheme (top left) with structures of unbound (top middle,

1109 PDB ID 2yb6) and Rad6 Binding Domain (R6BD) bound (top right, PDB ID 2ybf)

1110 Ube2B depicting the proposed gating and effector roles for Ube2B residues Glu58

1111 (thick outline) and Arg71 respectively. Green dashed and green lines depict Ube2B

1112 edges in unbound and R6BD bound states respectively. Grey lines depict common

1113 connections in the two Ube2B states while grey nodes have relative solvent

1114 accessibility of less than 10% in unbound E2. Immunoblots of an end-point (90 min)

1115 multi-turnover PCNA ubiquitination assay (bottom) shows Ube2B with a permissive

1116 Glu54Arg gate is more sensitive to increasing amounts of Rad18 E3 in substrate

1117 ubiquitination. Control immunoblots anti-Rad18 (middle) and anti-Ube2B (bottom)

1118 show levels of E3 and E2 respectively.

1119 B Representative time-course ubiquitination assays with fluorescently labelled ubiquitin

1120 (UbIR800) used to derive substrate ubiquitination rates depicted in Figure 5E. Substrate

1121 ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX).

1122 C Structures of human E2s with the proposed β3 gating residue (bold label), restrictive

1123 in red boxes while permissive in green boxes. Depicted below is a sequence alignment

1124 of human E2s, restrictive gate highlighted in red while permissive gate in green.

1125 Highlighted in grey is the proposed effector residue linked with the gate. Ube2T and

1126 Ube2B sequences are aligned as reference. Highlighted in grey is the proposed

1127 effector residue liked with the gate. Catalytic cysteine is also indicated (clear box).

1128 Molecular figures prepared in PyMOL (Schrödinger, LLC).

A C

FANCD2

DNA FANCI DNA FANCI

Wildtype K561R ← FANCD2 Wildtype K523R ← FANCI

0 4 8 16 32 32 (min) 0 4 8 16 32 32 (min)

250 - 250 - ← FANCD2 ← FANCI 150 - K561-Ub 150 - K523-Ub UbIR800 scan UbIR800 scan

150 - ← FANCD2 150 - ← FANCI ← FANCI Coomassie Coomassie

B D

UR RING-RING - ATP Ube2T - FANCL - ATP Ube2D3 - RNF4

Substrate Wildtype I44A Wildtype I44A ← Ubiquitin K523R K561R ← FANCI - FANCD2 - dsDNA FANCI - FANCD2 - dsDNA complex Ube2T 0 250 500 1000 0 250 500 1000 FANCLUR [nM] FANCD2 250 - K561-Ub + FANCI ← FANCD2 ← ubiquitination FANCIK523-Ub 150 - FANCD2 → UbIR800 scan K561-Ub

K561-Ub ← FANCD2 ← FANCD2 Wildtype K561R Wildtype K561R ← FANCD2 150 - ← FANCI (with FANCIWt)

Coomassie ← FANCD2-Ub 150 - ← FANCD2 ← FANCI

Coomassie

Figure 1. FANCLUR mediated FANCD2 ubiquitination does not require the ubiquitin Ile44-patch.

A. Time-course multi-turnover ubiquitination assays with fluorescently labelled ubiquitin (UbIR800) showing FANCLUR (0.1 μM) and Ube2T (0.1 μM) mediated site-specific mono-ubiquitination of Lys561 FANCD2 (1.0 μM) when present as a FANCD2-FANCI-dsDNA (1:1:2 μM) complex.

B. Titration of FANCLUR - Ube2T enzymes on FANCD2-FANCI-dsDNA (1:1:2 μM) complexes with single target lysine i.e. Lys561 FANCD2 or Lys523 FANCI reveals the FANCI site can be targeted but the FANCD2 site is preferred.

C. Time-course multi-turnover ubiquitination assays with fluorescently labelled ubiquitin (UbIR800) showing FANCLUR (0.1 μM) and Ube2T (0.1 μM) mediated site-specific mono-ubiquitination of Lys523 FANCI (1.0 μM) when present as a FANCI-dsDNA (1:2 μM) complex.

D. Comparing FANCD2-FANCI-dsDNA (1:1:2 μM) ubiquitination activities of the FANCLUR (0.1 μM) - Ube2T (0.1 μM) pair with RNF4RING-RING (0.1 μM) - Ube2D3 (0.2 μM) pair using fluorescently labelled wildtype and Ile44Ala ubiquitin. RNF4RING-RING and Ube2D3 robustly ubiquitinate FANCI-FANCD2 using wiltype ubiquitin, but is dramatically impaired by the Ile44Ala mutant while the FANCLUR and Ube2T maintain activity and site-specificity even with the ubiquitin mutant. Substrate ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX). bioRxiv preprint doi: https://doi.org/10.1101/429076; this version posted September 27, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A RING B helix1 L92A Q2 - ATP - E3 D33A D33A ∆ Wt Wt Wt R3A D32A D32A L92A L92-K95 ← Ube2T L7 250 - ← FANCD2 loop2 F63 150 - K561-Ub R3 L34 250 - D32 R60 ← FANCI R35 150 - K523-Ub UbIR800 scan D33

FANCL RING 4ccg RING C FANCD2 K561-Ub FANCI K523-Ub F63

helix1 1.0

L92 W98 loop2 K91 0.5 loop7 loop7 L90 Normalised activity loop8 K95 0.0 C86 Wt Wt R3A D32A D32A L92A ∆ N78 (- E3) D33A D33A L92-K95 Ube2T (1yh2) D122 L92A Ube2T (4ccg.A) Ube2T (4ccg.B) loop8 E Single-turnover FANCD2 ubiquitination

Ube2T1-152, K91R Ube2T1-152, K91R D32A, D33A, L92A

D 0 125 250 0 250125 500 1000 ← FANCLUR Ube2T1-152 autoubiquitination [nM] (E3 independent) 250 - ← FANCD2 0 8 16 32 0 8 16 32 0 8 16 32 (min) 150 - K561-Ub 25 - ← K91-Ub UbIR800 scan (reducing gel) 20 -

← E2 ∼ UbIR800 25 - 15 - Ube2T1-152 Ube2T1-152 Ube2T1-152 UbIR800 scan K91R D32A, D33A, L92A (non-reducing gel)

F FANCL URD-RING interaction ∆G ∆H -T∆S 8 Ube2T Wildtype Ube2T D32A, D33A, L92A 4 K = 116.5 ± 2.6 nM K = 121.5 ± 2.9 nM d d 0 N = 0.98 N = 0.94

-4 kcal/mol -8 -12 DP (µcal/s) DP (µcal/s) -16 Ube2T Ube2T Wildtype D32A, D33A, Time (min) Time (min) L92A

Figure 2. FANCL induced dynamics of Ube2T loop2 and loop7 is required for substrate ubiquitination.

A. Superpose of FANCLR (teal surface) bound copies of Ube2T (olive ribbon, PDB ID 4ccg.A and brown ribbon, PDB ID 4ccg.B) with the unbound Ube2T (grey ribbon) structure (PDB ID 1yh2) showing little overall structural change. Close-up of helix1-loop2 region (top) and loop7-loop8 region (bottom) reveal local changes. Molecular figures prepared in PyMOL (Schrödinger, LLC).

B. End-point (30 min) multi-turnover ubiquitination assay with fluorescently labelled ubiquitin (UbIR800) show conserved residues in Ube2T helix1, loop2 and loop7 are required for FANCLUR mediated FANCD2 and FANCI ubiquitination. Substrate ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX).

C. Effect of Ube2T helix1, loop2 and loop7 mutants on rates of FANCD2 and FANCI ubiquitination. A loop2-loop7 hybrid mutant (Asp32Ala, Asp33Ala, Leu92Ala) shows 75% loss in substrate ubiquitination rates. Rates normalized to wildtype levels and plotted as mean ± SD (n=3).

D. FANCL independent Ube2T1-152 autoubiquitination assay shows no effect of loop2-loop7 hybrid mutant in Lys91 autoubiquitination.

E. Single-turnover ubiquitination assay (10 min) of a FANCD2-FANCIK523R-dsDNA (2:2:2 μM) complex with increasing amounts of FANCLUR and 200 nM of Ube2T1-152, K91R ∼ UbIR800 thioester or Ube2T1-152, K91R loop2-loop7 hybrid mutant ∼ UbIR800 thioester shows the latter is defective in modifying Lys561 FANCD2.

F. Thermodynamics of FANCLUR interaction with Ube2T wildtype (left) and Ube2T loop2-loop7 hybrid mutant (middle) shows no change in binding affinity but divergent binding enthalpy. The cost of favourable enthalpy for the mutant is offset by reduced conformational entropy thus contributing to reduced activity. Graphs (right) plotted as mean ± range (n=2).

A B FANCL RING FANCL URD-RING interaction FANCL RING interaction

Ube2T Wt Ube2T T23R W25Q Ube2T Wt Ube2T T23R W25Q loop2 K = 116.5 ± 2.6 nM K = 200.0 ± 2.7 nM K = 249.5 ± 7.5 nM K = 259.0 ± 8.4 nM β1 d d d d W25

R35 T23 β2 Q37 ∆ H (kcal/mol) ∆ H (kcal/mol) V50 β3 L39

Molar Ratio Molar Ratio Ube2T Ube2T (RING bound) C D FANCD2 K561-Ub FANCI K523-Ub - ATP - E3 T23R Wt Wt Wt T23R W25Q Q37L ← Ube2T 1.0 250 - ← FANCD2 150 - K561-Ub 0.5 250 -

← FANCI Normalised activity 150 - K523-Ub 0.0 IR800 Ub scan Wt Wt T23R T23R Q37L (- E3) W25Q

E F Single-turnover FANCD2 ubiquitination Ube2T1-152 auto-ubiquitination Ube2T1-152, K91R Ube2T1-152, K91R T23R, W25Q (E3 independent) 0 125 250 0 250125 500 1000 ← FANCLUR 0 8 16 32 0 8 16 32 (min) [nM] 25 - 1-152 250 - ← Ube2T ← FANCD2 K91-Ub 20 - 150 - K561-Ub UbIR800 scan (reducing gel)

15 - 1-152 1-152 ← E2 ∼ UbIR800 Ube2T Ube2T 25 - T23R, W25Q UbIR800 scan (non-reducing gel)

Figure 3. Novel role of Ube2T backside in FANCL mediated substrate ubiquitination.

A. Superpose of FANCLR (teal surface) bound copy of Ube2T (olive ribbon, PDB ID 4ccg.A) with unbound Ube2T (grey ribbon) structure (PDB ID 1yh2) showing residues on E2 backside (β1 and β2) that are repositioned upon FANCL binding. Molecular figures prepared in PyMOL (Schrödinger, LLC).

B. Ube2T backside β1 double mutant (Thr23Arg, Trp25Gln) reveals binding defect with FANCLUR but is unaffected in FANCLR interaction. This suggests Ube2T backside supports additional FANCL interactions.

C. End-point (30 min) multi-turnover ubiquitination assay with fluorescently labelled ubiquitin (UbIR800) show Ube2T backside mutants are more defected in FANCLUR mediated FANCD2 ubiquitination than FANCI ubiquitination. Substrate ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX).

D. Effect of Ube2T backside mutants on the rates of FANCD2 and FANCI ubiquitination. Backside mutants reduce FANCI modification rate to levels observed in the no E3 setup. In contrast, a Ube2T backside β1 double mutant (Thr23Arg, Trp25Gln) shows 75% loss in rate of FANCD2 ubiquitination. Rates normalized to wildtype levels and plotted as mean ± SD (n=3).

E. FANCL independent Ube2T1-152 auto-ubiquitination assay shows no effect of backside β1 double mutant in Lys91 auto-ubiquitination.

F. Single-turnover ubiquitination assay (10 min) of a FANCD2-FANCIK523R-dsDNA (2:2:2 μM) complex with increasing amounts of FANCLUR and 200 nM of Ube2T1-152, K91R ∼ UbIR800 thioester or Ube2T1-152, K91R β1 mutant ∼ UbIR800 thioester shows the defect in the latter cannot be completely rescued by increasing FANCLUR levels. bioRxiv preprint doi: https://doi.org/10.1101/429076; this version posted September 27, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A B - ATP K95A Wt Wt R84S K91A D122A L124A ← Ube2T Ube2T allosteric 250 - network model ← FANCD2 RING cluster K561-Ub 150 -

S 5 F 63 250 - L 11 L 7 R 3 Q 2 ← FANCI backside cluster 150 - K523-Ub P 62 R 60 UbIR800 scan T 23 W 25 D 32 C K95A Wt R84S K91A D122A L124A Q 37 R 35 D 33 E 64 (ATP) L 34 +++++

← E2 � Ub

25 - ← E2 W 98 P 65

Coomassie L 53 K 52 E 54 L 92 D loop 7 )

-1 4 K 91 K 95 3 L 71 R 69 2 1 Reaction rate (pmoles.min 0 C 86 K91R K91R K91R K91A K91R ← Ube2T1-152 (no L-Lys) R84S K95A D122A G 83 R 84 catalytic β-element D 122 D 121 E A 82 K95R N 119 K95R K91R L 124 Wt K91A K95A K91R R84Q R84K R84K ← Ube2T S 81 P 123 P 120 free E2 D 80 250 - RING bound loop 8 ← FANCD2 M 125 K561-Ub free E2 RSA ≤ 10% 150 -

250 - ← FANCI 150 - K523-Ub UbIR800 scan

Figure 4. Allosteric residue network reveals Ube2t active-site residues critical for substrate ubiquitination.

A. Allosteric network model shows dynamic rewiring of Ube2T inter-residue connections upon FANCLR binding. Dashed and orange lines depict Ube2T edges in unbound and FANCLR bound state respectively. Grey nodes have relative solvent accessibility of less than 10% in unbound Ube2T. Nodes involved in RING binding are clustered in a cyan box while those predicted to support backside interaction are clustered in a purple box. Nodes in loop7 and loop8 are in unshaded boxes. Red node denotes the catalytic cysteine (Cys86) while the network termini nodes (Arg84, Lys91, Lys95, Asp122 and Leu124) that are within 10 Å of Cys86 have a thick outline. Also depicted is the catalytic β-element.

B. End-point (30 min) multi-turnover ubiquitination assay with fluorescently labelled ubiquitin (UbIR800) show mutations of the Ube2T network termini residues are detrimental to FANCLUR mediated FANCD2 and FANCI ubiquitination. Substrate ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX).

C. Ubiquitin charging assays of the network termini mutants show the Leu124Ala mutant alone is defected in E1-based E2 ∼ Ub thioester formation.

D. Lysine discharge assays show network termini mutants in loop7 (Lys91Ala + Lys95Ala) and loop8 (Asp122Ala) have catalytic defects while Arg84Ala is not defected. Graphs depict mean ± SD (n=2).

E. End-point (30 min) multi-turnover ubiquitination assay with fluorescently labelled ubiquitin (UbIR800) show the requirement of Arg/Lys residues in loop7 while Arg84, in the catalytic β-element, is critical for FANCI and FANCD2 ubiquitination. Partial compensation of activity for Ube2T Arg84Lys occurs only when the loop7 bears longer Arg residues. bioRxiv preprint doi: https://doi.org/10.1101/429076; this version posted September 27, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A B E54 - ATP Ube2T Wt Wt A Q DR E54 β3 E54 β3 E54 250 - K561-Ub FANCD2 ← VdW Ionic FANCIK523-Ub sc-sc sc-sc 150 - Hbond sc-mc β4 β4 250 -

R69 ← FANCI R69 R69 150 - K523-Ub

VdW IR800 mc-mc Ub scan Hbond G83 G83 sc-mc C

VdW G83 catalytic catalytic E54A with sc-mc β-element β-element - ATP Ube2T Wt Wt R69A R69K R69K R69A A82 R84 A82 R84 A82 250 - K561-Ub FANCD2 ← FANCIK523-Ub Free E2 RING bound C86 C86 150 - Free E2 RSA ≤ 10% 250 - Ube2T allosteric Ube2T RING - Ube2T ← FANCI conduit (1yh2) (4ccg.B) 150 - K523-Ub UbIR800 scan

D E FANCL FANCI K523-Ub FANCD2 K561-Ub R  32  

allosteric →  backside gate catalytic centre 24 Ube2T

RING allostery Backside binding RING binding mutant mutant mutant 16 (DDL/AAA) (TW/RQ) (F63A) - E3 Ube2T - ATP with with with with Wt Wt E54R E54R E54R E54R activityNormalised 8 250 - ← FANCD2 K561-Ub 150 - 0 UbIR800 scan Ube2T Ube2Tv1 Ube2Tv3

F G Ube2T - FANCLUR Ube2Tv3 - FANCLUR

0 1 0.125 0.25 0.5 1 1 [µM] FANCL 250 - RING ← FANCD2 150 - K561-Ub

RING backside K561R ← FANCD2 Wildtype [1µM]

UR UR Ube2T - FANCL Ube2Tv3 - FANCL loop7 loop7

0 1 0.125 0.25 0.5 1 1 [µM] allosteric allosteric gate Ub gate Ub 250 - catalytic centre ← FANCI 150 - K523-Ub Ube2T Ube2T FANCI / FANCD2 ubiquitination Wildtype K561R ← FANCI [1µM]

Figure 5. Ube2T deregulation leads to enhanced FANCL driven substrate ubiquitination.

A. Comparison of a network scheme (left) with structures of unbound (middle) and RING bound (right) Ube2T depicting the proposed allosteric conduit. A gating role is proposed for Glu54 (thick outline) for its regulation of Arg69. An effector role is proposed for Arg69 for its role in stabilising the catalytic β-element leading to a release of Arg84. Dashed and orange lines depict Ube2T edges in unbound and FANCLR bound state respectively. Grey nodes have relative solvent accessibility of less than 10% in unbound Ube2T.

B. End-point (30 min) multi-turnover ubiquitination assay with fluorescently labelled ubiquitin (UbIR800) shows Glu54 gating role is dependent on its negative side chain. Removal of this charge leads to improved FANCLUR mediated substrate ubiquitination. Substrate ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX).

C. End-point (30 min) multi-turnover ubiquitination assay with fluorescently labelled ubiquitin (UbIR800) shows effector role for Arg69 is linked to its positive side chain and is regulated by the Glu54 gate. While removal of the Glu54 negative charge improves improved FANCLUR mediated substrate ubiquitination, the removal of the Arg69 positive charge counters this effect.

D. End-point (30 min) multi-turnover ubiquitination assay of a FANCD2-FANCIK523R-dsDNA (1:1:2 μM) complex with fluorescently labelled ubiquitin (UbIR800) shows a permissive gate at position 54 (Glu54Arg) can rescue FANCD2 ubiquitination defects arising from the loop2-loop7 hybrid mutant (Asp32Ala, Asp33Ala, Leu92Ala), the backside β1 double mutant (Thr23Arg, Trp25Gln) as well as the RING binding mutant (Phe63Ala). Despite deregulation of the RING binding Ube2T Glu54Arg + Phe63Ala double mutant, FANCL is still required for efficient FANCD2 ubiquitination.

E. Effects of the Ube2Tv1 variant with a permissive gate (Glu54Arg) alongside the Ube2Tv3 variant containing both a permissive gate and flexible loop7 (Glu54Arg, Pro93Gly, Pro94Gly) on the rates of FANCLUR mediated FANCD2 (FANCIK523R-FANCD2-dsDNA complex) and FANCI (FANCI-dsDNA complex) ubiquitination.

F. End-point (30 min) multi-turnover ubiquitination assay of isolated FANCD2 and FANCI substrates with no DNA cofactors. Titration of FANCLUR - Ube2Tv3 enzymes shows how substrate ubiquitination is enhanced by the deregulated E2 without compromising on site-specificity.

G. Allosteric model for FANCLUR mediated activation of Ube2T. FANCL binding at classical RING-E2 interface and the backside E2 interface triggers long-range and short-range rewiring of Ube2T residue networks that culminate in substrate ubiquitination. The dynamic allosteric network is intrinsically regulated by conserved Ube2T elements, the allosteric β3 gating residue and a rigid loop7.

A B Zn PDB IDs 100 UR FANCL RNF4 172 TCPTCRK 178 4ap4 (30.3 kDa) BIRC7 281 LCPICRA 287 4auq AMFR 374 SCPTCRM 380 4lad RNF146 71 RCALCRQ 77 4qpl 75 RING1B 86 ECPTCRK 92 4s3o RNF38 499 TCPICRA 505 4v3l RNF165 330 KCPICRV 336 5d0k TRIM25 49 LCPQCRA 55 5fer 5mnj 50 MDM2 474 PCPVCRQ 480 FANCL 358 ECPYCSK 364 4ccg Ovalbumin (44.3 kDa) FANCL RNF4 RING RING UV 280 (mAU) BSA Carbonic 25 (66 kDa) A97 Q92 anhydrase S363 (29 kDa) R181

Q40 R72 0 Ube2T Ube2D1 Ubiquitin 4ccg 4ap4 5 10 15 20 Elution volume (ml)

C loop 2 α1 β1 β2 β3 β4

H. sapiens 1 MQRASRLKRELHMLATEPPPGITCWQ-DKDQMDDLRAQILGGANTPYEKGVFKLEVIIPERYPFEPPQIRFLTPIYH 76 B. taurus 1 MQRTSRLKRELSLLAAEPPPGITCWQ-DGDRMEDLRAQILGGANTPYEKGVFKLEVHIPERYPFEPPQIRFLTPIYH 76 R. norvegicus 1 MQRASRLKKELHMLAIEPPPGVTCWQ-EKDKMDNLRAQILGGANTPYEKGIFTLEVIVPERYPFEPPQIRFLTPIYH 76 O. anatinus 1 MQRASRLNKELKLLATEPPPGIMCWQ-ETSQMDELRAQILGGANTPYEKGIFKLEVVIPERYPFEPPKIRFLTPIYH 76 G. gallus 1 MQRASRLSRELTMLSTEPPPGISCWQ-SGARLDELRAQIIGAADTPYEKGIFDLEIVVPERYPFEPPKIRFLTPIYH 76 X. laevis 1 MQRVSRLKRELQLLNKEPPPGVICWQ-NESNMDDLRAQIIGGSGSPYEGGIFNLEIIVPERYPFEPPKIRFLTPIYHα1 76 D. rerio 1 MQRVSRLKREMQLLTAEPPPGVSCWQ-SEGRLDELQAQIVGGANTPYEGGVFTLEINIPERYPFEPPKMRFLTPIYH 76 N. vectensis 1 MQRNARMKRELRLLAEDPPHGASCWPVDENRIDKLEAKLIGAEGTPYHKGIFKLDIQIPERYPFEPPKVRFVTPIYH 77

loop 7 loop 8 310 α2 α3 α4

H. sapiens 77 PNIDSAGRICLDVLKLPPKGAWRPSLNIATVLTSIQLLMSEPNPDDPLMADISSEFKYNKPAFLKNARQWTEKHARQ 153 B. taurus 77 PNIDSAGRICLDVLKLPPKGAWRPSLNIATLLTSIQQLMAEPNPDDPLMADISSEFKYNKPVFFKNARQWTEKHARQ 153 R. norvegicus 77 PNIDSSGRICLDILKLPPKGAWRPSLNIATVLTSIQLLMAEPNPDDPLMADISSEFKYNKIAFVKKARQWTETHARQ 153 O. anatinus 77 PNIDSAGRICLDILKLPPKGAWRPSLNISTVLTSIQLLMSEPNSDDPLMADISSEFKYNKPAFLENARQWTEKYASQ 153 G. gallus 77 PNIDSAGRICLDVLKLPPKGAWRPSLNISTLLTSIQLLMVEPNPDDPLMADISSEYKYNKQLFLINAKEWTEKYASQ 153 X. laevis 77 PNIDSAGRICLDILKLPPKGAWRPALNISTVLTSIQLLMSEPNPDDPLMADISSEFKYNRAVFFSNAKKWTEKHALP 153 D. rerio 77 PNIDNAGRICLDALKLPPKGAWRPSLNISTVLTSIQLLMAEPNPDDPLMADISSEFKYNKPLYLEKAKKWTAEHAIQ 153 N. vectensis 78 PNIDSSGRICLDTLKMPPKGMWKPALNISSVLSTILILMAEPNPDDPLMAEISNEFKYNKAQFLEKAKEWTLKFAMD 154

Supplementary Figure 1. Solution and sequence profile of FANCLUR with Ube2T sequence conservation

A. Analytical size-exclusion chromatography of FANCLUR using Superdex75 10/300 GL column (blue trace). Gel-filtration standards (dashed trace) with known elution profile were also run in same buffer system as FANCLUR and the traces were overlaid. The FANCLUR fragment resolves as a monomeric protein however, the moderate shift to an earlier elution volume suggests the central UBC-RWD and C-terminal RING domain adopt an extended conformation as observed in the fly FANCL structure (PDB ID 3k1l).

B. Sequencing alignment of the C-terminal regions of various RING domains that have been characterised to have a linchpin Arg residue (blue highlight). FANCL appears to lack such a residue at the analogous position. Structure of the RNF4 linchpin residue (Arg181, green cartoon, PDB ID 4ap4) shows how the linchpin contacts both the E2 (Ube2D1, orange cartoon) and donor ubiquitin (yellow cartoon). In contrast, the equivalent FANCL residue Ser363 would not stabilise such conformation. Zinc co-ordinating cysteine residues are in bold. PDB IDs of respective structures are listed alongside. Molecular figures prepared in PyMOL (Schrödinger, LLC).

C. Sequence conservation of the UBC fold between Ube2T homologs. Residues shaded in red to yellow to highlight conservation, where red corresponds to strict conservation. Depicted above the sequences (grey) are secondary structure elements of human Ube2T are based on PDB ID 1yh2. Red star indicates catalytic cysteine while coloured circles indicate human Ube2T residues mutated in this study; blue in Figure 2, cyan in Figure 3, purple in Figure 4 and green in Figure 5. bioRxiv preprint doi: https://doi.org/10.1101/429076; this version posted September 27, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A B Ube2T ↓ 0 4 8 16 32 (min) 0 4 8 16 32 - E3 (Wt) L92A D33A D33A ∆ Wt R3A D32A D32A L92A L92-K95 Wt + + + + + + (ATP) 37 - R3A ← E2 � Ub D32A, D33A 25 - ← E2 D32A, D33A, L92A 20 -

L92A 10 - ← Ub ∆loop7

FANCD2 K561-Ub FANCI K523-Ub

C FANCL RING interaction ∆G ∆H -T∆S 8 Ube2T Wildtype Ube2T D32A, D33A, L92A 4

KD = 249.5 ± 7.5 nM KD = 251.5 ± 5.5 nM 0 N = 1.05

N = 0.95 kcal/mol -4 -8 DP (µcal/s) DP (µcal/s) -12 Ube2T Ube2T -16 Wildtype D32A, D33A, Time (min) Time (min) L92A

Supplementary Figure 2. Activity and binding profile of Ube2T loop2 and loop7 residues.

A. Representative time-course ubiquitination assays with fluorescently labelled ubiquitin (UbIR800) used to derive substrate ubiquitination rates depicted in Figure 2C. Substrate ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX).

B. Ubiquitin charging assays of the loop2 and loop7 mutants show that the mutants do not affect E1-based E2∼Ub thioester formation.

C. Thermodynamics of FANCLR interaction with Ube2T wildtype (left) and Ube2T loop2-loop7 hybrid mutant (middle) shows similar contrast in binding enthalpy as observed with FANCLUR. Graphs (right) plotted as mean ± range (n=2).

A B Ube2T T23R ↓ 0 4 8 16 32 (min) 0 4 8 16 32 Wt T23R W25Q Q37L - E3 (Wt) + + + + (ATP) 37 - Wt ← E2 � Ub T23R 25 - ← E2

T23R, W25Q 20 -

Q37L 10 - ← Ub FANCD2 K561-Ub FANCI K523-Ub

Supplementary Figure 3. Activity profile of Ube2T backside residues.

A. Representative time-course ubiquitination assays with fluorescently labelled ubiquitin (UbIR800) used to derive substrate ubiquitination rates depicted in Figure 3D. Substrate ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX).

B. Ubiquitin charging assays of Ube2T backside mutants show that the mutants do not affect E1-based E2∼Ub thioester formation. bioRxiv preprint doi: https://doi.org/10.1101/429076; this version posted September 27, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. A

D 29 K 28 T 16 K 48 E 17 P 18 N 43 T 23 L 39 P 19 Q 37 Q 30 A 42 A 15 RIN summary D 27 R 152 G 41 P 20 Q 26 H 12 W 25 V 50 G 21 M 13 Free Ube2T RIN (1yh2) → 152 nodes, 1467 edges E 47 G 49 T 72 L 14 M 31 A 151 I 22 C 24 D 33 L 11 K 8 R K 149 K 52 FANCL - Ube2T RIN (4ccg.A) → 152 nodes, 1454 edges E 148 G 40 A 36 T 147 T 44 I 38 R 9 H 150 R 35 L 34 D 32 FANCLR - Ube2T RIN (4ccg.B) → 146 nodes, 1368 edges A 105 A 4 F 51 Q 112 S 116 P 45 E 10 R 144 L 7 L 108 Q 145 Y 46 L 71 T 109 P 73 V 55 I 104 L 53 S 5 W 146 I 57 M 115 R 3 A 143 L 113 M 1 K 141 E 117 E 54 T 106 I 111 I 56 N 103 R 6 Comparison matrix summary I 74 L 140 F 70 Y 75 S 110 1yh2 edges 1yh2 edges N 142 V 107 P 58 F 139 H 76 P 66 E 59 Q 2 P 137 L 114 R 69 Y 61 A 138 P 118 I 68 92

119 L 102 K 136 I 79 Q 67 S 101 I 128 I 85 P 62 G 83 R 60 F 132 L 87 V 89 E 64 P 77 D 80 N 135 E 131 N 78 P 65 841 817 S 129 A 82 W 98 R 99 89 109 4ccg.A edges 4ccg.B edges S 81 D 88 L 90 R 84 K 133 M 125 F 63 D 127 A 97 G 96 P 100 D 121 L 124 Y 134 N 119 C 86 P 120 L 92 K 91 Free E2 S 130 D 122 shared A 126 P 123 Merged RIN P 93 RING bound K 95 P 94 Free E2 RSA ≤ 10%

B C D Ube2T1-152 Ube2T Arg84 network 0.1 1 2 4 (min) FANCL species R84 Free E2 RING D80 K91 ← E2 � Ub L 124 I 85 R84:L124 VdW (sc-sc) R84:D80 Hbond (sc-sc) C86 K91R L124 (no L-lysine) shared K95 ← E2 N78 R84:G83 VdW (sc-mc) D122 R84:D80 VdW (sc-sc) (1yh2) ← E2 � Ub R84:D80 Hbond (sc-mc) (4ccg.A) R84:D80 Ionic (sc-sc) (4ccg.B) K91R R 84 RING bound ← E2 R84:I85 Hbond (sc-mc) loop 7 loop 8 R84:A82 VdW (sc-mc) 310 ← E2 � Ub Ube2T YHPNIDS-AGRICLDVLKLP------PKGAW PNPD-D-PLM K91R B FHPNVYA-DGSICLDILQ------NRW PNPN-S-PAN R84S C YHPNVDT-QGNICLDILK------EKW PNID-S-PLN ← E2 D3 YHPNINS-NGSICLDILR------SQW PNPD-D-PLV E1 YHCNINS-QGVICLDILK------DNW CNPA-D-PLV ← E2 � Ub D 80 I 79 G2 FHPNIYP-DGRVCISILHAPGDDPMGYESSAERW PNDE-S-GAN K91A H FHPNIDEASGTVCLDVIN------QTW PNPI-D-PLN K95A K WHPNISSVTGAICLDILK------DQW AEPD-D-PQD ← E2 L3 YHPNIDE-KGQVCLPVIS------AENW PQPE-H-PLR ← E2 � Ub N YHPNVDK-LGRICLDILK------DKW PNPD-D-PLA K91R R1 WHPNIYE-TGDVCISILHPPVDDPQSGELPSERW PNTF-S-PAN D122A S FHPNVGA-NGEICVNVLK------RDW PNPE-S-ALN ← E2 W VHPHVYS-NGHICLSILT------EDW CKEKRRPPDN Coomassie A 82 G 83 * :: * :*: :: * : (IR700 scan)

FANCD2 homology

M. musculus 509 AAFVKGILDYLENM 522 548 HIQDDMHLVIRKQL 561 D552 H. sapiens 502 AVFVKGILDYLDNI 524 550 HIQDDMHLVIRKQL 563 S515 G. gallus 514 ATFVKTILDSMQKL 527 552 YIQDDMHMVIRKWL 565 D517 D551 X. laevis D485 513 AVFVKGILDYLDNL 526 551 HIQDDMFIVIRKQL 563 D514 D. rerio 510 TVFVKGILDYMDNL 525 550 HIQDDMHIVIRKQL 563 L518 L555 <10Å :.*** *** :::: :*****.:**** * <12Å

T488 FANCI homology K522 E520 K559 R558 R521 M. musculus 478 S-KVTETFDYLTFL 490 509 SMSMRDSLILVLRKA 523 H. sapiens 478 SSKVTEAFDYLSFL 491 510 SMSMRDCLILVLRKA 524 G. gallus 480 SSRVTETFDNLSFL 493 512 SMSVRDSLILVLQKA 526 X. laevis mouse FANCD2 479 SSKVTEAFDHLSFL 492 511 SMSMRDSLILVLRKA 525 mouse FANCI D. rerio (3s4w.A - 470 to 545) (3s4w.B - 501 to 585) 481 SSKVTETFDQLSYL 494 513 SMSMKDALILVLRKA 527 * :***:** *::** :*****.:**** **

Supplementary Figure 4. Ube2T residue network profile and characterization of network termini residues

A. Summary of nodes and edges (top left) in the individual Ube2T residue interaction networks (RINs). Mosaic plots of edge distribution profile derived from RIN comparison (bottom left). Complete profile of edges and nodes obtained after merging the RIN comparison networks. Dashed and orange lines depict Ube2T edges in unbound and FANCLR bound state respectively. Thin grey lines depict edges that are shared in both networks. Grey nodes have relative solvent accessibility of less than 10% in unbound Ube2T. Red node denotes the catalytic cysteine (Cys86).

B. Structural overlay (top) of unbound and RING bound Ube2T structures focussing on network termini residues surrounding the catalytic cysteine. Sequence alignment (bottom) of Ube2T catalytic site with various other human E2s reveal the termini of Ube2T network (indicated by boxes) vary among the enzyme family. Catalytic cysteine is shown in red.

C. Representative gels of lysine discharge assays plotted in Figure 4D. Coomassie stained gels are visualized by direct fluorescence monitoring in the red channel (Li-COR Odyssey CLX).

D. Total network profile of Arg84 residue present in Ube2T catalytic β-element. Listed alongside are edges that involve the Arg84 side chain. Edge depiction is same as in panel A.

E. Acidic/polar residues found proximal to target lysine on FANCD2 (left, green cartoon) and FANCI (right, blue cartoon) based on the mouse FANCI-FANCD2 complex structure (PDB ID 3s4w). Sequence alignment (middle) shows conservation of these acidic/polar residues (boxed). Respective target lysine is depicted in blue. bioRxiv preprint doi: https://doi.org/10.1101/429076; this version posted September 27, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. A - E2 Wt E58R ← Ube2B [2µM]

0 0 0.09 0.18 0.36 0.36 0 0.09 0.18 0.36 0.36 ← 6xHis-smt3Rad18 E58 β3 E58 β3 50 - [µM] VdW V56 sc-sc E58 V56 V56 ← 6xHisPCNA β4 β4 K164-Ub Hbond 37 - sc-sc T69 anti-PCNA R71 T69 R71 R71 Wt Wt ← 6xHisPCNA Hbond [9µM] sc-sc K164R K164R VdW sc-mc S86 catalytic S86 Hbond G85 S86 G85 catalytic 100 - sc-sc β-element Hbond G85 β-element 6xHis-smt3 sc-mc ← Rad18 75 - D84 C88 D84 Hbond anti-Rad18 D84 sc-sc C88 20 - Free E2 Ube2B R6BD - Ube2B Free E2 shared ← Ube2B RSA ≤ 10% induced by R6BD (2yb6) (2ybf) 15 - anti-Ube2B

Ube2T ↓ 0 2.5 5 10 (min) 0 2.5 5 10 Wt

Tv1

Tv3

FANCD2 K561-Ub FANCI K523-Ub

Ube2C β3 Ube2E1 β3 (4yii.U) E84 (5lbn.A)

D99 β4 β4

Ube2T GVFKLEVIIPERYPFEPPQIRFLTPIYHPNIDSA-GRICLDVLK S82 T114 A GTFKLTIEFTEEYPNKPPTVRFVSKMFHPNVYAD-GSICLDILQ B GTFKLVIEFSEEYPNKPPTVRFLSKMFHPNVYAD-GSICLDILQ K97 C LRYKLSLEFPSGYPYNAPTVKFLTPCYHPNVDTQ-GNICLDILK R116 catalytic catalytic E1 GVFFLDITFTPEYPFKPPKVTFRTRIYHCNINSQ-GVICLDILK β-element β-element E2 GVFFLDITFSPDYPFKPPKVTFRTRIYHCNINSQ-GVICLDILK N112 V129 E3 GVFFLDITFSSDYPFKPPKVTFRTRIYHCNINSQ-GVICLDILK C114 C131 L3 GAFRIEINFPAEYPFKPPKITFKTKIYHPNIDEK-GQVCLPVIS F GKFQFETEVPDAYNMVPPKVKCLTKIWHPNITET-GEICLSLLR K GRYQLEIKIPETYPFNPPKVRFITKIWHPNISSVTGAICLDILK β3 N GTFKLELFLPEEYPMAAPKVRFMTKIYHPNVDKL-GRICLDILK Ube2N Ube2G2 β3 G2 GVFPAILSFPLDYPLSPPKMRFTCEMFHPNIYPD-GRVCISILH (3W31.B) (4lad.A) I GLFKLRMLFKDDYPSSPPKCKFEPPLFHPNVYPS-GTVCLSILE E55 I57 β L6 KAFNLRISFPPEYPFKPPMIKFTTKIYHPNVDEN-GQICLPIIS 4 β4 M GKFQFSFKVGQGYPHDPPKVKCETMVYHPNIDLE-GNVCLNILR R1 GYFKARLKFPIDYPYSPPAFRFLTKMWHPNIYET-GDVCISILH : . : .* :* *: * :*: :: R70 R72 R87 catalytic catalytic R85 β-element β-element

C87 C89

Supplementary Figure 5. Activity profile of deregulated Ube2B and Ube2T variants.

A. Comparison of a network scheme (top left) with structures of unbound (top middle, PDB ID 2yb6) and Rad6 Binding Domain (R6BD) bound (top right, PDB ID 2ybf) Ube2B depicting the proposed gating and effector roles for Ube2B residues Glu58 (thick outline) and Arg71 respectively. Green dashed and green lines depict Ube2B edges in unbound and R6BD bound states respectively. Grey lines depict common connections in the two Ube2B states while grey nodes have relative solvent accessibility of less than 10% in unbound E2. Immunoblots of an end-point (90 min) multi-turnover PCNA ubiquitination assay (bottom) shows Ube2B with a permissive Glu54Arg gate is more sensitive to increasing amounts of Rad18 E3 in substrate ubiquitination. Control immunoblots anti-Rad18 (middle) and anti-Ube2B (bottom) show levels of E3 and E2 respectively.

B. Representative time-course ubiquitination assays with fluorescently labelled ubiquitin (UbIR800) used to derive substrate ubiquitination rates depicted in Figure 5E. Substrate ubiquitination is analysed by direct fluorescence monitoring (Li-COR Odyssey CLX).

C. Sequence alignment of human E2s, with proposed gating residues highlighted (restrictive in red and permissive in green). Highlighted in grey is the proposed effector residue linked with the gate. Ube2T and Ube2B sequences are aligned as reference. Catalytic cysteine is also indicated (clear box). Structures of indicated human E2s with the proposed β3 gating residue (bold label), restrictive in red boxes while permissive in green boxes. Molecular figures prepared in PyMOL (Schrödinger, LLC). bioRxiv preprint doi: https://doi.org/10.1101/429076; this version posted September 27, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Ube2T allosteric network net1 - free 2T network (Edges displayed in Fig. 4A) net2 - RING bound 2T network

Node 1 # Node 1 type Node 2 # Node 2 type Interaction Belongs To Comp Weight 2 Q 62 P VDW:MC_SC net2 -4.313 2 Q 62 P VDW:SC_SC net2 -4.043 2 Q 5 S VDW:MC_SC net1 4.337 2 Q 63 F VDW:SC_SC net2 0.221 2 Q 5 S HBOND:MC_SC net2 -3.12 3 R 32 D HBOND:SC_MC net2 -4.482 3 R 7 L VDW:MC_SC net2 -4.193 3 R 32 D VDW:SC_SC net1 0.212 3 R 32 D HBOND:SC_SC net1 4.697 3 R 32 D IONIC:SC_SC net1 4.442 3 R 5 S HBOND:MC_SC net2 -5.321 7 L 11 L VDW:MC_SC net2 -4.273 7 L 11 L VDW:SC_SC net2 0.364 7 L 34 L VDW:SC_SC net2 -4.119 11 L 25 W VDW:SC_MC net2 -4.202 23 T 25 W VDW:MC_SC net1 0.085 23 T 25 W VDW:SC_SC net2 0.399 25 W 35 R VDW:SC_MC net1 0.14 25 W 52 K VDW:SC_SC net1 4.194 32 D 34 L VDW:MC_SC net1 4.319 32 D 34 L HBOND:MC_MC net1 3.998 33 D 35 R IONIC:SC_SC net1 4.268 33 D 35 R HBOND:MC_SC net1 4.225 34 L 35 R HBOND:MC_SC net2 -4.915 35 R 53 L VDW:SC_MC net2 -4.226 35 R 52 K VDW:SC_SC net2 0.035 35 R 54 E HBOND:SC_SC net1 0.076 35 R 54 E VDW:SC_SC net1 4.206 35 R 53 L HBOND:SC_MC net2 -4.803 37 Q 52 K VDW:SC_SC net2 0.83 53 L 69 R VDW:SC_MC net1 0.055 54 E 59 R VDW:SC_SC net1 0.17 54 E 69 R HBOND:SC_MC net2 -5.205 54 E 69 R IONIC:SC_SC net1 3.786 60 R 63 F VDW:SC_SC net2 -3.941 60 R 63 F HBOND:SC_MC net2 -4.435 60 R 64 E IONIC:SC_SC net2 -4.912 60 R 63 F VDW:MC_SC net2 -0.119 62 P 64 E VDW:MC_MC net2 -4.31 62 P 98 W VDW:SC_SC net2 -4.334 63 F 98 W VDW:MC_SC net1 4.193 64 E 98 W HBOND:MC_SC net2 -3.799 65 P 91 K VDW:SC_MC net2 -4.229 69 R 82 A VDW:SC_MC net2 -4.33 69 R 83 G VDW:MC_MC net2 -4.332 69 R 83 G HBOND:SC_MC net2 -5.449 69 R 71 L VDW:SC_SC net1 4.073 80 D 83 G HBOND:SC_MC net2 -2.999 80 D 81 S HBOND:MC_SC net2 -4.785 80 D 81 S HBOND:SC_SC net1 4.98 80 D 81 S HBOND:SC_MC net1 4.32 80 D 82 A HBOND:MC_MC net1 3.093 80 D 84 R IONIC:SC_SC net1 4.943 80 D 84 R HBOND:SC_SC net1 3.481 80 D 124 L VDW:SC_SC net1 -0.044 81 S 125 M HBOND:SC_SC net2 -4.481 81 S 125 M HBOND:MC_SC net2 -4.03 82 A 84 R VDW:SC_SC net2 -3.889 84 R 124 L VDW:SC_SC net1 4.24 86 C 122 D VDW:SC_MC net2 -4.276 86 C 122 D VDW:SC_SC net2 0.371 86 C 123 P HBOND:SC_MC net1 5.474 91 K 92 L VDW:SC_MC net1 4.131 91 K 95 K VDW:SC_SC net1 -0.508 91 K 92 L HBOND:SC_MC net1 4.329 92 L 98 W HBOND:MC_MC net2 -5.424 bioRxiv preprint doi: https://doi.org/10.1101/429076; this version posted September 27, 2018. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

119 N 122 D HBOND:SC_SC net2 1.626 119 N 121 D VDW:MC_SC net1 3.816 119 N 120 P VDW:SC_MC net2 -4.047 119 N 120 P VDW:SC_SC net2 -0.679 119 N 121 D HBOND:MC_MC net2 -3.209 119 N 121 D HBOND:SC_MC net2 -2.972 119 N 122 D HBOND:SC_MC net2 -3.628 119 N 121 D VDW:SC_MC net2 -3.913 119 N 122 D VDW:SC_SC net2 -0.117 120 P 122 D HBOND:MC_MC net2 -3.718 121 D 122 D HBOND:SC_MC net1 3.904 122 D 123 P VDW:SC_SC net1 4.329 123 P 125 M VDW:SC_MC net1 4.2